diogo/yap-6.3
Archived
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

slipcover manual

Browse Source
This commit is contained in:
Fabrizio Riguzzi 2013-09-04 21:12:09 +02:00
parent 842b7f25ab
commit c671d8aa29
4 changed files with 358 additions and 194 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
packages/cplint/doc/bib.bib
View File

@ -1,3 +1,11 @@
@article{BelRig13-TPLP-IJ,
author = {Elena Bellodi and Fabrizio Riguzzi},
title = {Structure Learning of Probabilistic Logic Programs by Searching the Clause Space},
journal = {Theory and Practice of Logic Programming},
publisher = {Cambridge University Press},
copyright = {Cambridge University Press},
year = {2013}
}
@article{BelRig11-IDA, @article{BelRig11-IDA,
author = {Elena Bellodi and Fabrizio Riguzzi}, author = {Elena Bellodi and Fabrizio Riguzzi},
title = { Expectation {Maximization} over Binary Decision Diagrams for Probabilistic Logic Programs}, title = { Expectation {Maximization} over Binary Decision Diagrams for Probabilistic Logic Programs},

457
packages/cplint/doc/manual.html
View File

@ -7,7 +7,7 @@
<meta name="originator" content="TeX4ht (http://www.cse.ohio-state.edu/~gurari/TeX4ht/)"> <meta name="originator" content="TeX4ht (http://www.cse.ohio-state.edu/~gurari/TeX4ht/)">
<!-- html --> <!-- html -->
<meta name="src" content="manual.tex"> <meta name="src" content="manual.tex">
<meta name="date" content="2013-07-16 15:51:00"> <meta name="date" content="2013-09-04 21:08:00">
<link rel="stylesheet" type="text/css" href="manual.css"> <link rel="stylesheet" type="text/css" href="manual.css">
</head><body </head><body
> >
@ -21,18 +21,18 @@ class="cmr-12">Fabrizio Riguzzi</span>
<br /><span <br /><span
class="cmr-12">fabrizio.riguzzi@unife.it</span></div><br /> class="cmr-12">fabrizio.riguzzi@unife.it</span></div><br />
<div class="date" ><span <div class="date" ><span
class="cmr-12">July 16, 2013</span></div> class="cmr-12">September 4, 2013</span></div>
</div> </div>
<h3 class="sectionHead"><span class="titlemark">1 </span> <a <h3 class="sectionHead"><span class="titlemark">1 </span> <a
id="x1-10001"></a>Introduction</h3> id="x1-10001"></a>Introduction</h3>
<!--l. 31--><p class="noindent" ><span <!--l. 31--><p class="noindent" ><span
class="cmtt-10">cplint </span>is a suite of programs for reasoning with ICL <span class="cite">[<a class="cmtt-10">cplint </span>is a suite of programs for reasoning with ICL <span class="cite">[<a
href="#XDBLP:journals/ai/Poole97">13</a>]</span>, LPADs <span class="cite">[<a href="#XDBLP:journals/ai/Poole97">14</a>]</span>, LPADs <span class="cite">[<a
href="#XVenVer03-TR">22</a>,&#x00A0;<a href="#XVenVer03-TR">23</a>,&#x00A0;<a
href="#XVenVer04-ICLP04-IC">23</a>]</span> and href="#XVenVer04-ICLP04-IC">24</a>]</span> and
CP-logic programs <span class="cite">[<a CP-logic programs <span class="cite">[<a
href="#XVenDenBru-JELIA06">20</a>,&#x00A0;<a href="#XVenDenBru-JELIA06">21</a>,&#x00A0;<a
href="#XDBLP:journals/tplp/VennekensDB09">21</a>]</span>. It contains programs both for inference and href="#XDBLP:journals/tplp/VennekensDB09">22</a>]</span>. It contains programs both for inference and
learning. learning.
<!--l. 33--><p class="noindent" > <!--l. 33--><p class="noindent" >
<h3 class="sectionHead"><span class="titlemark">2 </span> <a <h3 class="sectionHead"><span class="titlemark">2 </span> <a
@ -121,7 +121,7 @@ h1:1&#x00A0;:-&#x00A0;b1,...,bm,\+&#x00A0;c1,...,\+&#x00A0;cl.
</div> </div>
<!--l. 83--><p class="nopar" > <!--l. 83--><p class="nopar" >
<!--l. 85--><p class="indent" > The coin example of <span class="cite">[<a <!--l. 85--><p class="indent" > The coin example of <span class="cite">[<a
href="#XVenVer04-ICLP04-IC">23</a>]</span> is represented as (see file <span href="#XVenVer04-ICLP04-IC">24</a>]</span> is represented as (see file <span
class="cmtt-10">coin.cpl</span>) class="cmtt-10">coin.cpl</span>)
<div class="verbatim" id="verbatim-7"> <div class="verbatim" id="verbatim-7">
@ -169,13 +169,13 @@ class="cmtt-10">cplint </span>contains various modules for answering queries.
<ul class="itemize1"> <ul class="itemize1">
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">lpadsld.pl</span>: uses the top-down procedure described in in <span class="cite">[<a class="cmtt-10">lpadsld.pl</span>: uses the top-down procedure described in in <span class="cite">[<a
href="#XRig-AIIA07-IC">14</a>]</span> and <span class="cite">[<a href="#XRig-AIIA07-IC">15</a>]</span> and <span class="cite">[<a
href="#XRig-RCRA07-IC">15</a>]</span>. It href="#XRig-RCRA07-IC">16</a>]</span>. It
is based on SLDNF resolution and is an adaptation of the interpreter for is based on SLDNF resolution and is an adaptation of the interpreter for
ProbLog <span class="cite">[<a ProbLog <span class="cite">[<a
href="#XDBLP:conf/ijcai/RaedtKT07">9</a>]</span>. href="#XDBLP:conf/ijcai/RaedtKT07">10</a>]</span>.
<!--l. 130--><p class="noindent" >It was proved correct <span class="cite">[<a <!--l. 130--><p class="noindent" >It was proved correct <span class="cite">[<a
href="#XRig-RCRA07-IC">15</a>]</span> with respect to the semantics of LPADs for href="#XRig-RCRA07-IC">16</a>]</span> with respect to the semantics of LPADs for
range restricted acyclic programs <span class="cite">[<a range restricted acyclic programs <span class="cite">[<a
href="#XDBLP:journals/ngc/AptB91">1</a>]</span> without function symbols. href="#XDBLP:journals/ngc/AptB91">1</a>]</span> without function symbols.
<!--l. 132--><p class="noindent" >It is also able to deal with extensions of LPADs and CP-logic: the clause <!--l. 132--><p class="noindent" >It is also able to deal with extensions of LPADs and CP-logic: the clause
@ -188,8 +188,8 @@ class="cmtt-10">setof </span>or <span
class="cmtt-10">bagof </span>operator. class="cmtt-10">bagof </span>operator.
These extended features have been introduced in order to represent These extended features have been introduced in order to represent
CLP(BN) <span class="cite">[<a CLP(BN) <span class="cite">[<a
href="#XSanPagQaz03-UAI-IC">19</a>]</span> programs and PRM models <span class="cite">[<a href="#XSanPagQaz03-UAI-IC">20</a>]</span> programs and PRM models <span class="cite">[<a
href="#XGetoor+al:JMLR02">12</a>]</span>: <span href="#XGetoor+al:JMLR02">13</a>]</span>: <span
class="cmtt-10">setof </span>and <span class="cmtt-10">setof </span>and <span
class="cmtt-10">bagof </span>allow to class="cmtt-10">bagof </span>allow to
express dependency of an attribute from an aggregate function of another express dependency of an attribute from an aggregate function of another
@ -199,11 +199,11 @@ class="cmtt-10">bagof </span>allow to
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">picl.pl</span>: performs inference on ICL programs <span class="cite">[<a class="cmtt-10">picl.pl</span>: performs inference on ICL programs <span class="cite">[<a
href="#XRig09-LJIGPL-IJ">16</a>]</span> href="#XRig09-LJIGPL-IJ">17</a>]</span>
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">lpad.pl</span>: uses a top-down procedure based on SLG resolution <span class="cite">[<a class="cmtt-10">lpad.pl</span>: uses a top-down procedure based on SLG resolution <span class="cite">[<a
href="#XDBLP:journals/jacm/ChenW96">8</a>]</span>. As a href="#XDBLP:journals/jacm/ChenW96">9</a>]</span>. As a
consequence, it works for any sound LPADs, i.e., any LPAD such that consequence, it works for any sound LPADs, i.e., any LPAD such that
each of its instances has a two valued well founded model. each of its instances has a two valued well founded model.
</li> </li>
@ -217,28 +217,28 @@ class="cmtt-10">cpl.pl</span>: uses a top-down procedure based on SLG resolution
<ul class="itemize2"> <ul class="itemize2">
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">deepit.pl </span>performs iterative deepening <span class="cite">[<a class="cmtt-10">deepit.pl </span>performs iterative deepening <span class="cite">[<a
href="#XBraRig10-ILP10-IC">7</a>]</span> href="#XBraRig10-ILP10-IC">8</a>]</span>
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">deepdyn.pl </span>performs dynamic iterative deepening <span class="cite">[<a class="cmtt-10">deepdyn.pl </span>performs dynamic iterative deepening <span class="cite">[<a
href="#XBraRig10-ILP10-IC">7</a>]</span> href="#XBraRig10-ILP10-IC">8</a>]</span>
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">bestk.pl </span>performs k-Best <span class="cite">[<a class="cmtt-10">bestk.pl </span>performs k-Best <span class="cite">[<a
href="#XBraRig10-ILP10-IC">7</a>]</span> href="#XBraRig10-ILP10-IC">8</a>]</span>
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">bestfirst.pl </span>performs best first <span class="cite">[<a class="cmtt-10">bestfirst.pl </span>performs best first <span class="cite">[<a
href="#XBraRig10-ILP10-IC">7</a>]</span> href="#XBraRig10-ILP10-IC">8</a>]</span>
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">montecarlo.pl </span>performs Monte Carlo <span class="cite">[<a class="cmtt-10">montecarlo.pl </span>performs Monte Carlo <span class="cite">[<a
href="#XBraRig10-ILP10-IC">7</a>]</span> href="#XBraRig10-ILP10-IC">8</a>]</span>
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">mcintyre.pl</span>: implements the algorithm MCINTYRE (Monte Carlo class="cmtt-10">mcintyre.pl</span>: implements the algorithm MCINTYRE (Monte Carlo
INference wiTh Yap REcord) <span class="cite">[<a INference wiTh Yap REcord) <span class="cite">[<a
href="#XRig11-CILC11-NC">17</a>]</span></li></ul> href="#XRig11-CILC11-NC">18</a>]</span></li></ul>
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">approx/exact.pl </span>as <span class="cmtt-10">approx/exact.pl </span>as <span
@ -276,7 +276,7 @@ class="cmmi-10">P</span>, i.e.,
sufficient causation, independent causation, no deus ex machina events sufficient causation, independent causation, no deus ex machina events
and temporal precedence. It uses the definition of the semantics given in and temporal precedence. It uses the definition of the semantics given in
<span class="cite">[<a <span class="cite">[<a
href="#XDBLP:journals/tplp/VennekensDB09">21</a>]</span>.</li></ul> href="#XDBLP:journals/tplp/VennekensDB09">22</a>]</span>.</li></ul>
<!--l. 159--><p class="noindent" > <!--l. 159--><p class="noindent" >
<h4 class="subsectionHead"><span class="titlemark">4.1 </span> <a <h4 class="subsectionHead"><span class="titlemark">4.1 </span> <a
id="x1-50004.1"></a>Commands</h4> id="x1-50004.1"></a>Commands</h4>
@ -827,55 +827,55 @@ class="cmtt-10">examples</span>:
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">alarm.cpl</span>: representation of the Bayesian network in Figure 2 of class="cmtt-10">alarm.cpl</span>: representation of the Bayesian network in Figure 2 of
<span class="cite">[<a <span class="cite">[<a
href="#XVenVer04-ICLP04-IC">23</a>]</span>. href="#XVenVer04-ICLP04-IC">24</a>]</span>.
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">coin.cpl</span>: coin example from <span class="cite">[<a class="cmtt-10">coin.cpl</span>: coin example from <span class="cite">[<a
href="#XVenVer04-ICLP04-IC">23</a>]</span>. href="#XVenVer04-ICLP04-IC">24</a>]</span>.
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">coin2.cpl</span>: coin example with two coins. class="cmtt-10">coin2.cpl</span>: coin example with two coins.
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">dice.cpl</span>: dice example from <span class="cite">[<a class="cmtt-10">dice.cpl</span>: dice example from <span class="cite">[<a
href="#XVenVer04-ICLP04-IC">23</a>]</span>. href="#XVenVer04-ICLP04-IC">24</a>]</span>.
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">twosideddice.cpl,</span><span class="cmtt-10">twosideddice.cpl,</span><span
class="cmtt-10">&#x00A0;threesideddice.cpl</span></span></span> game with idealized dice class="cmtt-10">&#x00A0;threesideddice.cpl</span></span></span> game with idealized dice
with two or three sides. Used in the experiments in <span class="cite">[<a with two or three sides. Used in the experiments in <span class="cite">[<a
href="#XRig-RCRA07-IC">15</a>]</span>. href="#XRig-RCRA07-IC">16</a>]</span>.
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">ex.cpl</span>: first example in <span class="cite">[<a class="cmtt-10">ex.cpl</span>: first example in <span class="cite">[<a
href="#XRig-RCRA07-IC">15</a>]</span>. href="#XRig-RCRA07-IC">16</a>]</span>.
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">exapprox.cpl</span>: example showing the problems of approximate class="cmtt-10">exapprox.cpl</span>: example showing the problems of approximate
inference (see <span class="cite">[<a inference (see <span class="cite">[<a
href="#XRig-RCRA07-IC">15</a>]</span>). href="#XRig-RCRA07-IC">16</a>]</span>).
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">exrange.cpl</span>: example showing the problems with non range class="cmtt-10">exrange.cpl</span>: example showing the problems with non range
restricted programs (see <span class="cite">[<a restricted programs (see <span class="cite">[<a
href="#XRig-RCRA07-IC">15</a>]</span>). href="#XRig-RCRA07-IC">16</a>]</span>).
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">female.cpl</span>: example showing the dependence of probabilities in the class="cmtt-10">female.cpl</span>: example showing the dependence of probabilities in the
head from variables in the body (from <span class="cite">[<a head from variables in the body (from <span class="cite">[<a
href="#XVenVer04-ICLP04-IC">23</a>]</span>). href="#XVenVer04-ICLP04-IC">24</a>]</span>).
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">mendel.cpl, mendels.cpl</span>: programs describing the Mendelian class="cmtt-10">mendel.cpl, mendels.cpl</span>: programs describing the Mendelian
rules of inheritance, taken from <span class="cite">[<a rules of inheritance, taken from <span class="cite">[<a
href="#XBlo04-ILP04WIP-IC">6</a>]</span>. href="#XBlo04-ILP04WIP-IC">7</a>]</span>.
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">paper_ref.cpl,</span><span class="cmtt-10">paper_ref.cpl,</span><span
class="cmtt-10">&#x00A0;paper_ref_simple.cpl</span></span></span>: paper citations examples, class="cmtt-10">&#x00A0;paper_ref_simple.cpl</span></span></span>: paper citations examples,
showing reference uncertainty, inspired by <span class="cite">[<a showing reference uncertainty, inspired by <span class="cite">[<a
href="#XGetoor+al:JMLR02">12</a>]</span>. href="#XGetoor+al:JMLR02">13</a>]</span>.
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">paper_ref_not.cpl</span></span></span>: paper citations example showing that negation class="cmtt-10">paper_ref_not.cpl</span></span></span>: paper citations example showing that negation
@ -895,13 +895,13 @@ class="cmtt-10">school.cpl</span>.
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">student.cpl</span></span></span>: student example from Figure 1.3 of <span class="cite">[<a class="cmtt-10">student.cpl</span></span></span>: student example from Figure 1.3 of <span class="cite">[<a
href="#XGetFri01-BC">11</a>]</span>. href="#XGetFri01-BC">12</a>]</span>.
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">win.cpl, light.cpl, trigger.cpl, throws.cpl, hiv.cpl,</span><br class="cmtt-10">win.cpl, light.cpl, trigger.cpl, throws.cpl, hiv.cpl,</span><br
class="newline" /> <span class="newline" /> <span
class="cmtt-10">invalid.cpl</span>: programs taken from <span class="cite">[<a class="cmtt-10">invalid.cpl</span>: programs taken from <span class="cite">[<a
href="#XDBLP:journals/tplp/VennekensDB09">21</a>]</span>. <span href="#XDBLP:journals/tplp/VennekensDB09">22</a>]</span>. <span
class="cmtt-10">invalid.cpl </span>is an example class="cmtt-10">invalid.cpl </span>is an example
of a program that is invalid but sound.</li></ul> of a program that is invalid but sound.</li></ul>
<!--l. 432--><p class="noindent" >The files <span <!--l. 432--><p class="noindent" >The files <span
@ -922,11 +922,11 @@ class="cmtt-10">cplint </span>contains the following learning algorithms:
class="cmtt-10">cplint </span>EM): an implementation of EM for learning parameters class="cmtt-10">cplint </span>EM): an implementation of EM for learning parameters
that is based on <span that is based on <span
class="cmtt-10">lpadsld.pl </span><span class="cite">[<a class="cmtt-10">lpadsld.pl </span><span class="cite">[<a
href="#XRigDiM11-ML-IJ">18</a>]</span> href="#XRigDiM11-ML-IJ">19</a>]</span>
</li> </li>
<li class="itemize">RIB (Relational Information Bottleneck): an algorithm for learning <li class="itemize">RIB (Relational Information Bottleneck): an algorithm for learning
parameters based on the Information Bottleneck <span class="cite">[<a parameters based on the Information Bottleneck <span class="cite">[<a
href="#XRigDiM11-ML-IJ">18</a>]</span> href="#XRigDiM11-ML-IJ">19</a>]</span>
</li> </li>
<li class="itemize">EMBLEM (EM over Bdds for probabilistic Logic programs Efficient <li class="itemize">EMBLEM (EM over Bdds for probabilistic Logic programs Efficient
Mining): an implementation of EM for learning parameters that computes Mining): an implementation of EM for learning parameters that computes
@ -935,14 +935,20 @@ href="#XBelRig11-IDA">5</a>,&#x00A0;<a
href="#XBelRig11-CILC11-NC">2</a>,&#x00A0;<a href="#XBelRig11-CILC11-NC">2</a>,&#x00A0;<a
href="#XBelRig11-TR">3</a>]</span> href="#XBelRig11-TR">3</a>]</span>
</li> </li>
<li class="itemize">SLIPCASE (Structure LearnIng of ProbabilistiC logic progrAmS with Em <li class="itemize">SLIPCASE (Structure LearnIng of ProbabilistiC logic progrAmS with
over bdds): an algorithm for learning the structure of program that is Em over bdds): an algorithm for learning the structure of programs by
based on EMBLEM <span class="cite">[<a searching directly the theory space <span class="cite">[<a
href="#XBelRig11-ILP11-IC">4</a>]</span></li></ul> href="#XBelRig11-ILP11-IC">4</a>]</span>
<!--l. 445--><p class="noindent" > </li>
<li class="itemize">SLIPCOVER (Structure LearnIng of Probabilistic logic programs by
searChing OVER the clause space): an algorithm for learning the structure
of programs by searching the clause space and the theory space separatery
<span class="cite">[<a
href="#XBelRig13-TPLP-IJ">6</a>]</span></li></ul>
<!--l. 446--><p class="noindent" >
<h4 class="subsectionHead"><span class="titlemark">5.1 </span> <a <h4 class="subsectionHead"><span class="titlemark">5.1 </span> <a
id="x1-110005.1"></a>Input</h4> id="x1-110005.1"></a>Input</h4>
<!--l. 446--><p class="noindent" >To execute the learning algorithms, prepare four files in the same folder: <!--l. 447--><p class="noindent" >To execute the learning algorithms, prepare four files in the same folder:
<ul class="itemize1"> <ul class="itemize1">
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">&#x003C;stem&#x003E;.kb</span>: contains the example interpretations class="cmtt-10">&#x003C;stem&#x003E;.kb</span>: contains the example interpretations
@ -956,11 +962,13 @@ class="cmtt-10">&#x003C;stem&#x003E;.l</span>: contains language bias informatio
</li> </li>
<li class="itemize"><span <li class="itemize"><span
class="cmtt-10">&#x003C;stem&#x003E;.cpl</span>: contains the LPAD for you which you want to learn the class="cmtt-10">&#x003C;stem&#x003E;.cpl</span>: contains the LPAD for you which you want to learn the
parameters or the initial LPAD for SLIPCASE</li></ul> parameters or the initial LPAD for SLIPCASE. For SLIPCOVER, this file
<!--l. 453--><p class="noindent" >where <span should be absent</li></ul>
<!--l. 454--><p class="noindent" >where <span
class="cmtt-10">&#x003C;stem&#x003E; </span>is your dataset name. Examples of these files can be found in the dataset class="cmtt-10">&#x003C;stem&#x003E; </span>is your dataset name. Examples of these files can be found in the dataset
pages. pages.
<!--l. 455--><p class="indent" > In <span <!--l. 456--><p class="indent" > In <span
class="cmtt-10">&#x003C;stem&#x003E;.kb </span>the example interpretations have to be given as a list of Prolog class="cmtt-10">&#x003C;stem&#x003E;.kb </span>the example interpretations have to be given as a list of Prolog
facts initiated by <span facts initiated by <span
class="cmtt-10">begin(model(&#x003C;name&#x003E;)). </span>and terminated by <span class="cmtt-10">begin(model(&#x003C;name&#x003E;)). </span>and terminated by <span
@ -980,27 +988,27 @@ begin(model(b1)).
&#x00A0;<br />actor(2). &#x00A0;<br />actor(2).
&#x00A0;<br />end(model(b1)). &#x00A0;<br />end(model(b1)).
</div> </div>
<!--l. 469--><p class="nopar" > The interpretations may contain a fact of the form <!--l. 470--><p class="nopar" > The interpretations may contain a fact of the form
<div class="verbatim" id="verbatim-37"> <div class="verbatim" id="verbatim-37">
prob(0.3). prob(0.3).
</div> </div>
<!--l. 473--><p class="nopar" > assigning a probability (0.3 in this case) to the interpretations. If this is omitted, the <!--l. 474--><p class="nopar" > assigning a probability (0.3 in this case) to the interpretations. If this is omitted, the
probability of each interpretation is considered equal to 1<span probability of each interpretation is considered equal to 1<span
class="cmmi-10">&#x2215;n </span>where <span class="cmmi-10">&#x2215;n </span>where <span
class="cmmi-10">n </span>is the total class="cmmi-10">n </span>is the total
number of interpretations. <span class="obeylines-h"><span class="verb"><span number of interpretations. <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">prob/1</span></span></span> can be used to set different multiplicity for the class="cmtt-10">prob/1</span></span></span> can be used to set different multiplicity for the
different interpretations. different interpretations.
<!--l. 476--><p class="indent" > In order for RIB to work, the input interpretations must share the Herbrand <!--l. 477--><p class="indent" > In order for RIB to work, the input interpretations must share the Herbrand
universe. If this is not the case, you have to translate the interpretations in this was, universe. If this is not the case, you have to translate the interpretations in this was,
see for example the <span see for example the <span
class="cmtt-10">sp1 </span>files in RIB&#8217;s folder, that are the results of the conversion of class="cmtt-10">sp1 </span>files in RIB&#8217;s folder, that are the results of the conversion of
the first fold of the IMDB dataset. the first fold of the IMDB dataset.
<!--l. 478--><p class="indent" > <span <!--l. 479--><p class="indent" > <span
class="cmtt-10">&#x003C;stem&#x003E;.bg </span>can contain Prolog clauses that can be used to derive additional class="cmtt-10">&#x003C;stem&#x003E;.bg </span>can contain Prolog clauses that can be used to derive additional
conclusions from the atoms in the interpretations. conclusions from the atoms in the interpretations.
<!--l. 481--><p class="indent" > <span <!--l. 482--><p class="indent" > <span
class="cmtt-10">&#x003C;stem&#x003E;.l </span>contains the declarations of the input and output predicates, of the class="cmtt-10">&#x003C;stem&#x003E;.l </span>contains the declarations of the input and output predicates, of the
unseen predicates and the commands for setting the algorithms&#8217; parameters. Output unseen predicates and the commands for setting the algorithms&#8217; parameters. Output
predicates are declared as predicates are declared as
@ -1008,72 +1016,92 @@ predicates are declared as
<div class="verbatim" id="verbatim-38"> <div class="verbatim" id="verbatim-38">
output(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;). output(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;).
</div> </div>
<!--l. 485--><p class="nopar" > and define the predicates whose atoms in the input interpretations are used as the <!--l. 486--><p class="nopar" > and define the predicates whose atoms in the input interpretations are used as the
goals for the prediction of which you want to optimize the parameters. Derivations goals for the prediction of which you want to optimize the parameters. Derivations
for these goals are built by the systems. for these goals are built by the systems.
<!--l. 488--><p class="indent" > Input predicates are those for the predictions of which you do not want to <!--l. 489--><p class="indent" > Input predicates are those for the predictions of which you do not want to
optimize the parameters. You can declare closed world input predicates optimize the parameters. You can declare closed world input predicates
with with
<div class="verbatim" id="verbatim-39"> <div class="verbatim" id="verbatim-39">
input_cw(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;). input_cw(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;).
</div> </div>
<!--l. 491--><p class="nopar" > For these predicates, the only true atoms are those in the interpretations, the <!--l. 492--><p class="nopar" > For these predicates, the only true atoms are those in the interpretations, the
clauses in the input program are not used to derive atoms not present in the clauses in the input program are not used to derive atoms not present in the
interpretations. interpretations.
<!--l. 494--><p class="indent" > Open world input predicates are declared with <!--l. 495--><p class="indent" > Open world input predicates are declared with
<div class="verbatim" id="verbatim-40"> <div class="verbatim" id="verbatim-40">
input(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;). input(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;).
</div> </div>
<!--l. 497--><p class="nopar" > In this case, if a subgoal for such a predicate is encountered when deriving the atoms <!--l. 498--><p class="nopar" > In this case, if a subgoal for such a predicate is encountered when deriving the atoms
for the output predicates, both the facts in the interpretations and the clauses of the for the output predicates, both the facts in the interpretations and the clauses of the
input program are used. input program are used.
<!--l. 501--><p class="indent" > For RIB, if there are unseen predicates, i.e., predicates that are present in the <!--l. 502--><p class="indent" > For RIB, if there are unseen predicates, i.e., predicates that are present in the
input program but not in the interpretations, you have to declare them input program but not in the interpretations, you have to declare them
with with
<div class="verbatim" id="verbatim-41"> <div class="verbatim" id="verbatim-41">
unseen(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;). unseen(&#x003C;predicate&#x003E;/&#x003C;arity&#x003E;).
</div> </div>
<!--l. 504--><p class="nopar" > <!--l. 505--><p class="nopar" >
<!--l. 506--><p class="indent" > For SLIPCASE, you have to specify the language bias by means of mode <!--l. 507--><p class="indent" > For SLIPCASE and SLIPCOVER, you have to specify the language bias by
declarations in the style of <a means of mode declarations in the style of <a
href="http://www.doc.ic.ac.uk/~shm/progol.html" > Progol </a>. href="http://www.doc.ic.ac.uk/~shm/progol.html" > Progol </a>.
<div class="verbatim" id="verbatim-42"> <div class="verbatim" id="verbatim-42">
modeh(&#x003C;recall&#x003E;,&#x003C;predicate&#x003E;(&#x003C;arg1&#x003E;,...). modeh(&#x003C;recall&#x003E;,&#x003C;predicate&#x003E;(&#x003C;arg1&#x003E;,...).
</div> </div>
<!--l. 510--><p class="nopar" > specifies the atoms that can appear in the head of clauses, while <!--l. 511--><p class="nopar" > specifies the atoms that can appear in the head of clauses, while
<div class="verbatim" id="verbatim-43"> <div class="verbatim" id="verbatim-43">
modeb(&#x003C;recall&#x003E;,&#x003C;predicate&#x003E;(&#x003C;arg1&#x003E;,...). modeb(&#x003C;recall&#x003E;,&#x003C;predicate&#x003E;(&#x003C;arg1&#x003E;,...).
</div> </div>
<!--l. 514--><p class="nopar" > specifies the atoms that can appear in the body of clauses. <span <!--l. 515--><p class="nopar" > specifies the atoms that can appear in the body of clauses. <span
class="cmtt-10">&#x003C;recall&#x003E; </span>can be an class="cmtt-10">&#x003C;recall&#x003E; </span>can be an
integer or <span integer or <span
class="cmtt-10">* </span>(currently unused). class="cmtt-10">* </span>(currently unused).
<!--l. 518--><p class="indent" > The arguments are of the form <!--l. 519--><p class="indent" > The arguments are of the form
<div class="verbatim" id="verbatim-44"> <div class="verbatim" id="verbatim-44">
+&#x003C;type&#x003E; +&#x003C;type&#x003E;
</div> </div>
<!--l. 521--><p class="nopar" > for specifying an input variable of type <span <!--l. 522--><p class="nopar" > for specifying an input variable of type <span
class="cmtt-10">&#x003C;type&#x003E;</span>, or class="cmtt-10">&#x003C;type&#x003E;</span>, or
<div class="verbatim" id="verbatim-45"> <div class="verbatim" id="verbatim-45">
-&#x003C;type&#x003E; -&#x003C;type&#x003E;
</div> </div>
<!--l. 525--><p class="nopar" > for specifying an output variable of type <span <!--l. 526--><p class="nopar" > for specifying an output variable of type <span
class="cmtt-10">&#x003C;type&#x003E;</span>. or class="cmtt-10">&#x003C;type&#x003E;</span>. or
<div class="verbatim" id="verbatim-46"> <div class="verbatim" id="verbatim-46">
&#x003C;constant&#x003E; &#x003C;constant&#x003E;
</div> </div>
<!--l. 529--><p class="nopar" > for specifying a constant. <!--l. 530--><p class="nopar" > for specifying a constant.
<!--l. 532--><p class="indent" > An example of language bias for the UWCSE domain is <!--l. 533--><p class="indent" > SLIPCOVER also allows the arguments
<div class="verbatim" id="verbatim-47"> <div class="verbatim" id="verbatim-47">
#&#x003C;type&#x003E;
</div>
<!--l. 536--><p class="nopar" > for specifying an argument which should be replaced by a constant of type <span
class="cmtt-10">&#x003C;type&#x003E; </span>in
the bottom clause but should not be used for replacing input variables of the
following literals or
<div class="verbatim" id="verbatim-48">
-#&#x003C;type&#x003E;
</div>
<!--l. 540--><p class="nopar" > for specifying an argument which should be replaced by a constant of type <span
class="cmtt-10">&#x003C;type&#x003E; </span>in
the bottom clause and that should be used for replacing input variables of
the following literals. <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">#</span></span></span> and <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">-#</span></span></span> differ only in the creation of the bottom
clause.
<!--l. 543--><p class="indent" > An example of language bias for the UWCSE domain is
<div class="verbatim" id="verbatim-49">
output(advisedby/2). output(advisedby/2).
&#x00A0;<br /> &#x00A0;<br />
&#x00A0;<br />input(student/1). &#x00A0;<br />input(student/1).
@ -1090,18 +1118,57 @@ output(advisedby/2).
&#x00A0;<br />modeb(*,samecourse(-course,&#x00A0;+course)). &#x00A0;<br />modeb(*,samecourse(-course,&#x00A0;+course)).
&#x00A0;<br />.... &#x00A0;<br />....
</div> </div>
<!--l. 549--><p class="nopar" > <!--l. 560--><p class="nopar" > SLIPCOVER also requires facts for the <span class="obeylines-h"><span class="verb"><span
<!--l. 551--><p class="noindent" > class="cmtt-10">determination/2</span></span></span> predicate that indicate
which predicates can appear in the body of clauses. For example
<div class="verbatim" id="verbatim-50">
determination(professor/1,student/1).
&#x00A0;<br />determination(student/1,hasposition/2).
</div>
<!--l. 566--><p class="nopar" > state that <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">student/1</span></span></span> can appear in the body of clauses for <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">professor/1</span></span></span> and that
<span class="obeylines-h"><span class="verb"><span
class="cmtt-10">hasposition/2</span></span></span> can appear in the body of clauses for <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">student/1</span></span></span>.
<!--l. 570--><p class="indent" > SLIPCOVER also allows mode declarations of the form
<div class="verbatim" id="verbatim-51">
modeh(&#x003C;r&#x003E;,[&#x003C;s1&#x003E;,...,&#x003C;sn&#x003E;],[&#x003C;a1&#x003E;,...,&#x003C;an&#x003E;],[&#x003C;P1/Ar1&#x003E;,...,&#x003C;Pk/Ark&#x003E;]).
</div>
<!--l. 573--><p class="nopar" > These mode declarations are used to generate clauses with more than two head
atoms. In them, <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">&#x003C;s1&#x003E;,...,&#x003C;sn&#x003E;</span></span></span> are schemas, <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">&#x003C;a1&#x003E;,...,&#x003C;an&#x003E;</span></span></span> are atoms such that
<span class="obeylines-h"><span class="verb"><span
class="cmtt-10">&#x003C;ai&#x003E;</span></span></span> is obtained from <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">&#x003C;si&#x003E;</span></span></span> by replacing placemarkers with variables, <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">&#x003C;Pi/Ari&#x003E;</span></span></span> are
the predicates admitted in the body. <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">&#x003C;a1&#x003E;,...,&#x003C;an&#x003E;</span></span></span> are used to indicate which
variables should be shared by the atoms in the head. An example of such a mode
declaration is
<div class="verbatim" id="verbatim-52">
modeh(*,
&#x00A0;<br />&#x00A0;&#x00A0;[advisedby(+person,+person),tempadvisedby(+person,+person)],
&#x00A0;<br />&#x00A0;&#x00A0;[advisedby(A,B),tempadvisedby(A,B)],
&#x00A0;<br />&#x00A0;&#x00A0;[professor/1,student/1,hasposition/2,inphase/2,
&#x00A0;<br />&#x00A0;&#x00A0;publication/2,taughtby/3,ta/3,courselevel/2,yearsinprogram/2]).
</div>
<!--l. 583--><p class="nopar" >
<!--l. 587--><p class="noindent" >
<h4 class="subsectionHead"><span class="titlemark">5.2 </span> <a <h4 class="subsectionHead"><span class="titlemark">5.2 </span> <a
id="x1-120005.2"></a>Parameters</h4> id="x1-120005.2"></a>Parameters</h4>
<!--l. 552--><p class="noindent" >In order to set the algorithms&#8217; parameters, you have to insert in <span <!--l. 588--><p class="noindent" >In order to set the algorithms&#8217; parameters, you have to insert in <span
class="cmtt-10">&#x003C;stem&#x003E;.l </span>commands class="cmtt-10">&#x003C;stem&#x003E;.l </span>commands
of the form of the form
<div class="verbatim" id="verbatim-48"> <div class="verbatim" id="verbatim-53">
:-&#x00A0;set(&#x003C;parameter&#x003E;,&#x003C;value&#x003E;). :-&#x00A0;set(&#x003C;parameter&#x003E;,&#x003C;value&#x003E;).
</div> </div>
<!--l. 555--><p class="nopar" > The available parameters are: <!--l. 591--><p class="nopar" > The available parameters are:
<ul class="itemize1"> <ul class="itemize1">
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">depth</span></span></span> (values: integer or <span class="obeylines-h"><span class="verb"><span class="cmtt-10">depth</span></span></span> (values: integer or <span class="obeylines-h"><span class="verb"><span
@ -1114,10 +1181,10 @@ class="cmtt-10">true</span></span></span>
class="cmtt-10">single_var</span></span></span> (values: <span class="obeylines-h"><span class="verb"><span class="cmtt-10">single_var</span></span></span> (values: <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">{true,false}</span></span></span>, default value: <span class="obeylines-h"><span class="verb"><span class="cmtt-10">{true,false}</span></span></span>, default value: <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">false</span></span></span>, valid for CEM, class="cmtt-10">false</span></span></span>, valid for CEM,
EMBLEM and SLIPCASE): if set to <span class="obeylines-h"><span class="verb"><span EMBLEM, SLIPCASE and SLIPCOVER): if set to <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">true</span></span></span>, there is a random variable for class="cmtt-10">true</span></span></span>, there is a
each clauses, instead of a separate random variable for each grounding of random variable for each clauses, instead of a separate random variable
a clause for each grounding of a clause
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">sample_size</span></span></span> (values: integer, default value: 1000): total number of class="cmtt-10">sample_size</span></span></span> (values: integer, default value: 1000): total number of
@ -1128,37 +1195,38 @@ class="cmtt-10">prob(P).</span></span></span>
class="cmtt-10">sample_size*P</span></span></span> examples class="cmtt-10">sample_size*P</span></span></span> examples
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">epsilon_em</span></span></span> (values: real, default value: 0.1, valid for CEM, EMBLEM class="cmtt-10">epsilon_em</span></span></span> (values: real, default value: 0.1, valid for CEM, EMBLEM,
and SLIPCASE): if the difference in the log likelihood in two successive SLIPCASE and SLIPCOVER): if the difference in the log likelihood in
EM iteration is smaller than <span class="obeylines-h"><span class="verb"><span two successive EM iteration is smaller than <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">epsilon_em</span></span></span>, then EM stops class="cmtt-10">epsilon_em</span></span></span>, then EM stops
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">epsilon_em_fraction</span></span></span> (values: real, default value: 0.01, valid for CEM, class="cmtt-10">epsilon_em_fraction</span></span></span> (values: real, default value: 0.01, valid for
EMBLEM and SLIPCASE): if the difference in the log likelihood in two CEM, EMBLEM, SLIPCASE and SLIPCOVER): if the difference in
successive EM iteration is smaller than <span class="obeylines-h"><span class="verb"><span the log likelihood in two successive EM iteration is smaller than
class="cmtt-10">epsilon_em_fraction</span></span></span>*(-current <span class="obeylines-h"><span class="verb"><span
log likelihood), then EM stops class="cmtt-10">epsilon_em_fraction</span></span></span>*(-current log likelihood), then EM stops
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">iter</span></span></span> (values: integer, defualt value: 1, valid for EMBLEM and class="cmtt-10">iter</span></span></span> (values: integer, defualt value: 1, valid for EMBLEM, SLIPCASE and
SLIPCASE): maximum number of iteration of EM parameter learning. If SLIPCOVER): maximum number of iteration of EM parameter learning.
set to -1, no maximum number of iterations is imposed If set to -1, no maximum number of iterations is imposed
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">iterREF</span></span></span> (values: integer, defualt value: 1, valid for SLIPCASE): maximum class="cmtt-10">iterREF</span></span></span> (values: integer, defualt value: 1, valid for SLIPCASE and
number of iteration of EM parameter learning for refinements. If set to -1, SLIPCOVER): maximum number of iteration of EM parameter learning
no maximum number of iterations is imposed. for refinements. If set to -1, no maximum number of iterations is imposed.
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">random_restarts_number</span></span></span> (values: integer, default value: 1, valid for class="cmtt-10">random_restarts_number</span></span></span> (values: integer, default value: 1, valid for
CEM, EMBLEM and SLIPCASE): number of random restarts of EM CEM, EMBLEM, SLIPCASE and SLIPCOVER): number of random
learning restarts of EM learning
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">random_restarts_REFnumber</span></span></span> (values: integer, default value: 1, valid for class="cmtt-10">random_restarts_REFnumber</span></span></span> (values: integer, default value: 1, valid for
SLIPCASE): number of random restarts of EM learning for refinements SLIPCASE and SLIPCOVER): number of random restarts of EM learning
for refinements
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">setrand</span></span></span> (values: rand(integer,integer,integer)): seed for the random class="cmtt-10">setrand</span></span></span> (values: rand(integer,integer,integer)): seed for the random
@ -1185,7 +1253,7 @@ class="cmtt-10">logsize_fraction</span></span></span> times its maximum
class="cmsy-10">|</span><span class="cmsy-10">|</span><span
class="cmmi-10">CH,T</span><span class="cmmi-10">CH,T</span><span
class="cmsy-10">|</span>, see <span class="cite">[<a class="cmsy-10">|</span>, see <span class="cite">[<a
href="#XDBLP:journals/jmlr/ElidanF05">10</a>]</span>) href="#XDBLP:journals/jmlr/ElidanF05">11</a>]</span>)
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">delta</span></span></span> (values: negative integer, default value -10, valid for RIB): value class="cmtt-10">delta</span></span></span> (values: negative integer, default value -10, valid for RIB): value
@ -1195,7 +1263,7 @@ class="cmtt-10">delta</span></span></span> (values: negative integer, default va
class="cmtt-10">epsilon_fraction</span></span></span> (values: integer, default value 100, valid for RIB): class="cmtt-10">epsilon_fraction</span></span></span> (values: integer, default value 100, valid for RIB):
in the computation of the step, the value of <span in the computation of the step, the value of <span
class="cmmi-10">&#x03F5; </span>of <span class="cite">[<a class="cmmi-10">&#x03F5; </span>of <span class="cite">[<a
href="#XDBLP:journals/jmlr/ElidanF05">10</a>]</span> is obtained as href="#XDBLP:journals/jmlr/ElidanF05">11</a>]</span> is obtained as
log <span log <span
class="cmsy-10">|</span><span class="cmsy-10">|</span><span
class="cmmi-10">CH,T</span><span class="cmmi-10">CH,T</span><span
@ -1211,7 +1279,7 @@ class="cmtt-10">max_rules</span></span></span> (values: integer, default valu
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">logzero</span></span></span> (values: negative real, default value log(0<span class="cmtt-10">logzero</span></span></span> (values: negative real, default value log(0<span
class="cmmi-10">.</span>000001), valid for class="cmmi-10">.</span>000001), valid for
SLIPCASE): value assigned to log 0 SLIPCASE and SLIPCOVER): value assigned to log 0
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">examples</span></span></span> (values: <span class="obeylines-h"><span class="verb"><span class="cmtt-10">examples</span></span></span> (values: <span class="obeylines-h"><span class="verb"><span
@ -1233,94 +1301,125 @@ class="cmtt-10">group</span></span></span> (values: integer, default value: 1, v
target atoms in the groups that are used to build BDDs target atoms in the groups that are used to build BDDs
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">nax_iter</span></span></span> (values: integer, default value: 10, valid for SLIPCASE): number class="cmtt-10">nax_iter</span></span></span> (values: integer, default value: 10, valid for SLIPCASE and
of interations of beam search SLIPCOVER): number of interations of beam search
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">max_var</span></span></span> (values: integer, default value: 1, valid for SLIPCASE): maximum class="cmtt-10">max_var</span></span></span> (values: integer, default value: 1, valid for SLIPCASE and
number of distinct variables in a clause SLIPCOVER): maximum number of distinct variables in a clause
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">verbosity</span></span></span> (values: integer in [1,3], default value: 1): level of verbosity of class="cmtt-10">verbosity</span></span></span> (values: integer in [1,3], default value: 1): level of verbosity of
the algorithms the algorithms
</li> </li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">beamsize</span></span></span> (values: integer, default value: 20, valid for SLIPCASE): size of class="cmtt-10">beamsize</span></span></span> (values: integer, default value: 20, valid for SLIPCASE and
the beam in SLIPCASE SLIPCOVER): size of the beam
</li></ul> </li>
<!--l. 590--><p class="noindent" > <li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">megaex_bottom</span></span></span> (values: integer, default value: 1, valid for SLIPCOVER):
number of mega-examples on which to build the bottom clauses
</li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">initial_clauses_per_megaex</span></span></span> (values: integer, default value: 1, valid for
SLIPCOVER): number of bottom clauses to build for each mega-example
</li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">d</span></span></span> (values: integer, default value: 10000, valid for SLIPCOVER): number
of saturation steps when building the bottom clause
</li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">max_iter_structure</span></span></span> (values: integer, default value: 1, valid for
SLIPCOVER): maximum number of theory search iterations
</li>
<li class="itemize"><span class="obeylines-h"><span class="verb"><span
class="cmtt-10">background_clauses</span></span></span> (values: integer, default value: 50, valid for
SLIPCOVER): maximum numbers of background clauses.</li></ul>
<!--l. 631--><p class="noindent" >
<h4 class="subsectionHead"><span class="titlemark">5.3 </span> <a <h4 class="subsectionHead"><span class="titlemark">5.3 </span> <a
id="x1-130005.3"></a>Commands</h4> id="x1-130005.3"></a>Commands</h4>
<!--l. 591--><p class="noindent" >To execute CEM, load <span <!--l. 632--><p class="noindent" >To execute CEM, load <span
class="cmtt-10">em.pl </span>with class="cmtt-10">em.pl </span>with
<div class="verbatim" id="verbatim-49"> <div class="verbatim" id="verbatim-54">
?:-&#x00A0;use_module(library(&#8217;cplint/em&#8217;)). ?:-&#x00A0;use_module(library(&#8217;cplint/em&#8217;)).
</div> </div>
<!--l. 594--><p class="nopar" > and call: <!--l. 635--><p class="nopar" > and call:
<div class="verbatim" id="verbatim-50">
?:-&#x00A0;em(stem).
</div>
<!--l. 598--><p class="nopar" > To execute RIB, load <span
class="cmtt-10">rib.pl </span>with
<div class="verbatim" id="verbatim-51">
?:-&#x00A0;use_module(library(&#8217;cplint/rib&#8217;)).
</div>
<!--l. 602--><p class="nopar" > and call:
<div class="verbatim" id="verbatim-52">
?:-&#x00A0;ib_par(stem).
</div>
<!--l. 606--><p class="nopar" > To execute EMBLEM, load <span
class="cmtt-10">slipcase.pl </span>with
<div class="verbatim" id="verbatim-53">
?:-&#x00A0;use_module(library(&#8217;cplint/slipcase&#8217;)).
</div>
<!--l. 610--><p class="nopar" > and call
<div class="verbatim" id="verbatim-54">
?:-&#x00A0;em(stem).
</div>
<!--l. 614--><p class="nopar" > To execute SLIPCASE, load <span
class="cmtt-10">slipcase.pl </span>with
<div class="verbatim" id="verbatim-55"> <div class="verbatim" id="verbatim-55">
?:-&#x00A0;use_module(library(&#8217;cplint/slipcase&#8217;)). ?:-&#x00A0;em(stem).
</div> </div>
<!--l. 618--><p class="nopar" > and call <!--l. 639--><p class="nopar" > To execute RIB, load <span
class="cmtt-10">rib.pl </span>with
<div class="verbatim" id="verbatim-56"> <div class="verbatim" id="verbatim-56">
?:-&#x00A0;use_module(library(&#8217;cplint/rib&#8217;)).
</div>
<!--l. 643--><p class="nopar" > and call:
<div class="verbatim" id="verbatim-57">
?:-&#x00A0;ib_par(stem).
</div>
<!--l. 647--><p class="nopar" > To execute EMBLEM, load <span
class="cmtt-10">slipcase.pl </span>with
<div class="verbatim" id="verbatim-58">
?:-&#x00A0;use_module(library(&#8217;cplint/slipcase&#8217;)).
</div>
<!--l. 651--><p class="nopar" > and call
<div class="verbatim" id="verbatim-59">
?:-&#x00A0;em(stem).
</div>
<!--l. 655--><p class="nopar" > To execute SLIPCASE, load <span
class="cmtt-10">slipcase.pl </span>with
<div class="verbatim" id="verbatim-60">
?:-&#x00A0;use_module(library(&#8217;cplint/slipcase&#8217;)).
</div>
<!--l. 659--><p class="nopar" > and call
<div class="verbatim" id="verbatim-61">
?:-&#x00A0;sl(stem). ?:-&#x00A0;sl(stem).
</div> </div>
<!--l. 622--><p class="nopar" > <!--l. 663--><p class="nopar" > To execute SLIPCOVER, load <span
class="cmtt-10">slipcover.pl </span>with
<div class="verbatim" id="verbatim-62">
?:-&#x00A0;use_module(library(&#8217;cplint/slipcover&#8217;)).
</div>
<!--l. 667--><p class="nopar" > and call
<div class="verbatim" id="verbatim-63">
?:-&#x00A0;sl(stem).
</div>
<!--l. 671--><p class="nopar" >
<h4 class="subsectionHead"><span class="titlemark">5.4 </span> <a <h4 class="subsectionHead"><span class="titlemark">5.4 </span> <a
id="x1-140005.4"></a>Learning Examples</h4> id="x1-140005.4"></a>Learning Examples</h4>
<!--l. 624--><p class="noindent" >The subfolders <span class="obeylines-h"><span class="verb"><span <!--l. 673--><p class="noindent" >The subfolders <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">em</span></span></span>, <span class="obeylines-h"><span class="verb"><span class="cmtt-10">em</span></span></span>, <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">rib</span></span></span> and <span class="obeylines-h"><span class="verb"><span class="cmtt-10">rib</span></span></span>, <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">slipcase</span></span></span> of the <span class="obeylines-h"><span class="verb"><span class="cmtt-10">slipcase</span></span></span> and <span class="obeylines-h"><span class="verb"><span
class="cmtt-10">packages/cplint</span></span></span> folder in Yap class="cmtt-10">slipcover</span></span></span> of the <span class="obeylines-h"><span class="verb"><span
git distribution contain examples of input and output files for the learning class="cmtt-10">packages/cplint</span></span></span> folder in
Yap git distribution contain examples of input and output files for the learning
algorithms. algorithms.
<!--l. 627--><p class="noindent" > <!--l. 676--><p class="noindent" >
<h3 class="sectionHead"><span class="titlemark">6 </span> <a <h3 class="sectionHead"><span class="titlemark">6 </span> <a
id="x1-150006"></a>License</h3> id="x1-150006"></a>License</h3>
<!--l. 632--><p class="noindent" ><span <!--l. 681--><p class="noindent" ><span
class="cmtt-10">cplint</span>, as Yap, follows the Artistic License 2.0 that you can find in Yap CVS root class="cmtt-10">cplint</span>, as Yap, follows the Artistic License 2.0 that you can find in Yap CVS root
dir. The copyright is by Fabrizio Riguzzi. dir. The copyright is by Fabrizio Riguzzi.
<!--l. 635--><p class="indent" > The modules in the approx subdirectory use SimplecuddLPADs, a modification of <!--l. 684--><p class="indent" > The modules in the approx subdirectory use SimplecuddLPADs, a modification of
the <a the <a
href="http://dtai.cs.kuleuven.be/problog/download.html" > Simplecudd </a> library whose copyright is by Katholieke Universiteit Leuven and href="http://dtai.cs.kuleuven.be/problog/download.html" > Simplecudd </a> library whose copyright is by Katholieke Universiteit Leuven and
that follows the Artistic License 2.0. that follows the Artistic License 2.0.
<!--l. 638--><p class="indent" > Some modules use the library <a <!--l. 687--><p class="indent" > Some modules use the library <a
href="http://vlsi.colorado.edu/~fabio/" > CUDD </a> for manipulating BDDs that is included in href="http://vlsi.colorado.edu/~fabio/" > CUDD </a> for manipulating BDDs that is included in
glu. For the use of CUDD, the following license must be accepted: glu. For the use of CUDD, the following license must be accepted:
<!--l. 643--><p class="indent" > Copyright (c) 1995-2004, Regents of the University of Colorado <!--l. 692--><p class="indent" > Copyright (c) 1995-2004, Regents of the University of Colorado
<!--l. 645--><p class="indent" > All rights reserved. <!--l. 694--><p class="indent" > All rights reserved.
<!--l. 647--><p class="indent" > Redistribution and use in source and binary forms, with or without modification, <!--l. 696--><p class="indent" > Redistribution and use in source and binary forms, with or without modification,
are permitted provided that the following conditions are met: are permitted provided that the following conditions are met:
<ul class="itemize1"> <ul class="itemize1">
<li class="itemize">Redistributions of source code must retain the above copyright notice, this <li class="itemize">Redistributions of source code must retain the above copyright notice, this
@ -1334,7 +1433,7 @@ are permitted provided that the following conditions are met:
contributors may be used to endorse or promote products derived from contributors may be used to endorse or promote products derived from
this software without specific prior written permission.</li></ul> this software without specific prior written permission.</li></ul>
<!--l. 664--><p class="noindent" >THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS <br <!--l. 713--><p class="noindent" >THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS <br
class="newline" />AND CONTRIBUTORS &#8221;AS IS&#8221; AND ANY EXPRESS OR IMPLIED class="newline" />AND CONTRIBUTORS &#8221;AS IS&#8221; AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
@ -1348,7 +1447,7 @@ class="newline" />AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
<!--l. 678--><p class="indent" > <span <!--l. 727--><p class="indent" > <span
class="cmtt-10">lpad.pl</span>, <span class="cmtt-10">lpad.pl</span>, <span
class="cmtt-10">semlpad.pl </span>and <span class="cmtt-10">semlpad.pl </span>and <span
class="cmtt-10">cpl.pl </span>are based on the SLG system by Weidong class="cmtt-10">cpl.pl </span>are based on the SLG system by Weidong
@ -1402,6 +1501,14 @@ class="cmti-10">Intel. Data Anal.</span>,
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[6]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [6]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XBelRig13-TPLP-IJ"></a>Elena Bellodi and Fabrizio Riguzzi. Structure learning of probabilistic
logic programs by searching the clause space. <span
class="cmti-10">Theory and Practice of Logic</span>
<span
class="cmti-10">Programming</span>, 2013.
</p>
<p class="bibitem" ><span class="biblabel">
[7]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XBlo04-ILP04WIP-IC"></a>H.&#x00A0;Blockeel. Probabilistic logical models for mendel&#8217;s experiments: An id="XBlo04-ILP04WIP-IC"></a>H.&#x00A0;Blockeel. Probabilistic logical models for mendel&#8217;s experiments: An
exercise. In <span exercise. In <span
class="cmti-10">Inductive Logic Programming (ILP 2004), Work in Progress</span> class="cmti-10">Inductive Logic Programming (ILP 2004), Work in Progress</span>
@ -1409,7 +1516,7 @@ class="cmti-10">Inductive Logic Programming (ILP 2004), Work in Progress</span>
class="cmti-10">Track</span>, 2004. class="cmti-10">Track</span>, 2004.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[7]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [8]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XBraRig10-ILP10-IC"></a>Stefano Bragaglia and Fabrizio Riguzzi. Approximate inference for logic id="XBraRig10-ILP10-IC"></a>Stefano Bragaglia and Fabrizio Riguzzi. Approximate inference for logic
programs with annotated disjunctions. In Paolo Frasconi and Francesca programs with annotated disjunctions. In Paolo Frasconi and Francesca
Lisi, editors, <span Lisi, editors, <span
@ -1420,13 +1527,13 @@ class="cmti-10">ILP 2010, Florence, Italy, June 27-30, 2010. Revised Papers</spa
class="cmti-10">LNCS</span>, pages 30&#8211;37. Springer, 2011. class="cmti-10">LNCS</span>, pages 30&#8211;37. Springer, 2011.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[8]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [9]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XDBLP:journals/jacm/ChenW96"></a>Weidong Chen and David&#x00A0;Scott Warren. Tabled evaluation with id="XDBLP:journals/jacm/ChenW96"></a>Weidong Chen and David&#x00A0;Scott Warren. Tabled evaluation with
delaying for general logic programs. <span delaying for general logic programs. <span
class="cmti-10">Journal of the ACM</span>, 43(1):20&#8211;74, 1996. class="cmti-10">Journal of the ACM</span>, 43(1):20&#8211;74, 1996.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[9]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [10]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XDBLP:conf/ijcai/RaedtKT07"></a>L.&#x00A0;De&#x00A0;Raedt, A.&#x00A0;Kimmig, and H.&#x00A0;Toivonen. ProbLog: A probabilistic id="XDBLP:conf/ijcai/RaedtKT07"></a>L.&#x00A0;De&#x00A0;Raedt, A.&#x00A0;Kimmig, and H.&#x00A0;Toivonen. ProbLog: A probabilistic
Prolog and its application in link discovery. In <span Prolog and its application in link discovery. In <span
class="cmti-10">International Joint</span> class="cmti-10">International Joint</span>
@ -1434,22 +1541,22 @@ class="cmti-10">International Joint</span>
class="cmti-10">Conference on Artificial Intelligence</span>, pages 2462&#8211;2467, 2007. class="cmti-10">Conference on Artificial Intelligence</span>, pages 2462&#8211;2467, 2007.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[10]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [11]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XDBLP:journals/jmlr/ElidanF05"></a>G.&#x00A0;Elidan and N.&#x00A0;Friedman. Learning hidden variable networks: The id="XDBLP:journals/jmlr/ElidanF05"></a>G.&#x00A0;Elidan and N.&#x00A0;Friedman. Learning hidden variable networks: The
information bottleneck approach. <span information bottleneck approach. <span
class="cmti-10">Journal of Machine Learning Research</span>, class="cmti-10">Journal of Machine Learning Research</span>,
6:81&#8211;127, 2005. 6:81&#8211;127, 2005.
</p>
<p class="bibitem" ><span class="biblabel">
[11]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XGetFri01-BC"></a>L.&#x00A0;Getoor, N.&#x00A0;Friedman, D.&#x00A0;Koller, and A.&#x00A0;Pfeffer. Learning
probabilistic relational models. In Saso Dzeroski and Nada Lavrac, editors,
<span
class="cmti-10">Relational Data Mining</span>. Springer-Verlag, Berlin, 2001.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[12]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [12]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XGetFri01-BC"></a>L.&#x00A0;Getoor, N.&#x00A0;Friedman, D.&#x00A0;Koller, and A.&#x00A0;Pfeffer. Learning
probabilistic relational models. In Saso Dzeroski and Nada Lavrac, editors,
<span
class="cmti-10">Relational Data Mining</span>. Springer-Verlag, Berlin, 2001.
</p>
<p class="bibitem" ><span class="biblabel">
[13]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XGetoor+al:JMLR02"></a>L.&#x00A0;Getoor, N.&#x00A0;Friedman, D.&#x00A0;Koller, and B.&#x00A0;Taskar. Learning id="XGetoor+al:JMLR02"></a>L.&#x00A0;Getoor, N.&#x00A0;Friedman, D.&#x00A0;Koller, and B.&#x00A0;Taskar. Learning
probabilistic models of relational structure. <span probabilistic models of relational structure. <span
class="cmti-10">Journal of Machine Learning</span> class="cmti-10">Journal of Machine Learning</span>
@ -1457,13 +1564,13 @@ class="cmti-10">Journal of Machine Learning</span>
class="cmti-10">Research</span>, 3:679&#8211;707, December 2002. class="cmti-10">Research</span>, 3:679&#8211;707, December 2002.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[13]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [14]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XDBLP:journals/ai/Poole97"></a>David Poole. The independent choice logic for modelling multiple agents id="XDBLP:journals/ai/Poole97"></a>David Poole. The independent choice logic for modelling multiple agents
under uncertainty. <span under uncertainty. <span
class="cmti-10">Artificial Intelligence</span>, 94(1-2):7&#8211;56, 1997. class="cmti-10">Artificial Intelligence</span>, 94(1-2):7&#8211;56, 1997.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[14]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [15]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XRig-AIIA07-IC"></a>Fabrizio Riguzzi. A top down interpreter for LPAD and CP-logic. In id="XRig-AIIA07-IC"></a>Fabrizio Riguzzi. A top down interpreter for LPAD and CP-logic. In
<span <span
class="cmti-10">Congress of the Italian Association for Artificial Intelligence</span>, volume 4733 class="cmti-10">Congress of the Italian Association for Artificial Intelligence</span>, volume 4733
@ -1471,7 +1578,7 @@ class="cmti-10">Congress of the Italian Association for Artificial Intelligence<
class="cmti-10">LNAI</span>, pages 109&#8211;120. Springer, 2007. class="cmti-10">LNAI</span>, pages 109&#8211;120. Springer, 2007.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[15]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [16]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XRig-RCRA07-IC"></a>Fabrizio Riguzzi. A top down interpreter for LPAD and CP-logic. id="XRig-RCRA07-IC"></a>Fabrizio Riguzzi. A top down interpreter for LPAD and CP-logic.
In <span In <span
class="cmti-10">Proceedings of the 14th RCRA workshop Experimental Evaluation of</span> class="cmti-10">Proceedings of the 14th RCRA workshop Experimental Evaluation of</span>
@ -1479,13 +1586,13 @@ class="cmti-10">Proceedings of the 14th RCRA workshop Experimental Evaluation of
class="cmti-10">Algorithms for Solving Problems with Combinatorial Explosion</span>, 2007. class="cmti-10">Algorithms for Solving Problems with Combinatorial Explosion</span>, 2007.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[16]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [17]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XRig09-LJIGPL-IJ"></a>Fabrizio Riguzzi. Extended semantics and inference for the Independent id="XRig09-LJIGPL-IJ"></a>Fabrizio Riguzzi. Extended semantics and inference for the Independent
Choice Logic. <span Choice Logic. <span
class="cmti-10">Logic Journal of the IGPL</span>, 17(6):589&#8211;629, 2009. class="cmti-10">Logic Journal of the IGPL</span>, 17(6):589&#8211;629, 2009.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[17]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [18]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XRig11-CILC11-NC"></a>Fabrizio Riguzzi. MCINTYRE: A Monte Carlo algorithm for id="XRig11-CILC11-NC"></a>Fabrizio Riguzzi. MCINTYRE: A Monte Carlo algorithm for
probabilistic logic programming. In <span probabilistic logic programming. In <span
class="cmti-10">Proceedings of the 26th Italian</span> class="cmti-10">Proceedings of the 26th Italian</span>
@ -1495,14 +1602,15 @@ class="cmti-10">Conference on Computational Logic (CILC2011), Pescara, Ita
class="cmti-10">August-2 September, 2011</span>, 2011. class="cmti-10">August-2 September, 2011</span>, 2011.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[18]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [19]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XRigDiM11-ML-IJ"></a>Fabrizio Riguzzi and Nicola&#x00A0;Di Mauro. Applying the information id="XRigDiM11-ML-IJ"></a>Fabrizio Riguzzi and Nicola&#x00A0;Di Mauro. Applying the information
bottleneck to statistical relational learning. <span bottleneck to statistical relational learning. <span
class="cmti-10">Machine Learning</span>, 2011. To class="cmti-10">Machine Learning</span>, 2011. To
appear. appear.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[19]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [20]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XSanPagQaz03-UAI-IC"></a>V.&#x00A0;Santos&#x00A0;Costa, D.&#x00A0;Page, M.&#x00A0;Qazi, and J.&#x00A0;Cussens. CLP(<span id="XSanPagQaz03-UAI-IC"></a>V.&#x00A0;Santos&#x00A0;Costa, D.&#x00A0;Page, M.&#x00A0;Qazi, and J.&#x00A0;Cussens. CLP(<span
class="cmsy-10"><img class="cmsy-10"><img
src="cmsy10-42.png" alt="B" class="10x-x-42" /><img src="cmsy10-42.png" alt="B" class="10x-x-42" /><img
@ -1511,10 +1619,9 @@ src="cmsy10-4e.png" alt="N" class="10x-x-4e" /></span>):
class="cmti-10">Uncertainty</span> class="cmti-10">Uncertainty</span>
<span <span
class="cmti-10">in Artificial Intelligence</span>. Morgan Kaufmann, 2003. class="cmti-10">in Artificial Intelligence</span>. Morgan Kaufmann, 2003.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[20]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [21]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XVenDenBru-JELIA06"></a>J.&#x00A0;Vennekens, M.&#x00A0;Denecker, and M.&#x00A0;Bruynooghe. Representing causal id="XVenDenBru-JELIA06"></a>J.&#x00A0;Vennekens, M.&#x00A0;Denecker, and M.&#x00A0;Bruynooghe. Representing causal
information about a probabilistic process. In <span information about a probabilistic process. In <span
class="cmti-10">Proceedings of the 10th</span> class="cmti-10">Proceedings of the 10th</span>
@ -1523,19 +1630,19 @@ class="cmti-10">European Conference on Logics in Artificial Intelligence</span>,
September 2006. September 2006.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[21]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [22]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XDBLP:journals/tplp/VennekensDB09"></a>J.&#x00A0;Vennekens, Marc Denecker, and Maurice Bruynooghe. CP-logic: id="XDBLP:journals/tplp/VennekensDB09"></a>J.&#x00A0;Vennekens, Marc Denecker, and Maurice Bruynooghe. CP-logic:
A language of causal probabilistic events and its relation to logic A language of causal probabilistic events and its relation to logic
programming. <span programming. <span
class="cmti-10">Theory Pract. Log. Program.</span>, 9(3):245&#8211;308, 2009. class="cmti-10">Theory Pract. Log. Program.</span>, 9(3):245&#8211;308, 2009.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[22]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [23]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XVenVer03-TR"></a>J.&#x00A0;Vennekens and S.&#x00A0;Verbaeten. Logic programs with annotated id="XVenVer03-TR"></a>J.&#x00A0;Vennekens and S.&#x00A0;Verbaeten. Logic programs with annotated
disjunctions. Technical Report CW386, K. U. Leuven, 2003. disjunctions. Technical Report CW386, K. U. Leuven, 2003.
</p> </p>
<p class="bibitem" ><span class="biblabel"> <p class="bibitem" ><span class="biblabel">
[23]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a [24]<span class="bibsp">&#x00A0;&#x00A0;&#x00A0;</span></span><a
id="XVenVer04-ICLP04-IC"></a>J.&#x00A0;Vennekens, S.&#x00A0;Verbaeten, and M.&#x00A0;Bruynooghe. Logic programs id="XVenVer04-ICLP04-IC"></a>J.&#x00A0;Vennekens, S.&#x00A0;Verbaeten, and M.&#x00A0;Bruynooghe. Logic programs
with annotated disjunctions. In <span with annotated disjunctions. In <span
class="cmti-10">International Conference on Logic</span> class="cmti-10">International Conference on Logic</span>

BIN
packages/cplint/doc/manual.pdf
View File

Binary file not shown.

87
packages/cplint/doc/manual.tex
View File

@ -439,7 +439,8 @@ The files \texttt{*.uni} that are present for some of the examples are used by
\item CEM (\texttt{cplint} EM): an implementation of EM for learning parameters that is based on \texttt{lpadsld.pl} \cite{RigDiM11-ML-IJ} \item CEM (\texttt{cplint} EM): an implementation of EM for learning parameters that is based on \texttt{lpadsld.pl} \cite{RigDiM11-ML-IJ}
\item RIB (Relational Information Bottleneck): an algorithm for learning parameters based on the Information Bottleneck \cite{RigDiM11-ML-IJ} \item RIB (Relational Information Bottleneck): an algorithm for learning parameters based on the Information Bottleneck \cite{RigDiM11-ML-IJ}
\item EMBLEM (EM over Bdds for probabilistic Logic programs Efficient Mining): an implementation of EM for learning parameters that computes expectations directly on BDDs \cite{BelRig11-IDA,BelRig11-CILC11-NC,BelRig11-TR} \item EMBLEM (EM over Bdds for probabilistic Logic programs Efficient Mining): an implementation of EM for learning parameters that computes expectations directly on BDDs \cite{BelRig11-IDA,BelRig11-CILC11-NC,BelRig11-TR}
\item SLIPCASE (Structure LearnIng of ProbabilistiC logic progrAmS with Em over bdds): an algorithm for learning the structure of program that is based on EMBLEM \cite{BelRig11-ILP11-IC} \item SLIPCASE (Structure LearnIng of ProbabilistiC logic progrAmS with Em over bdds): an algorithm for learning the structure of programs by searching directly the theory space \cite{BelRig11-ILP11-IC}
\item SLIPCOVER (Structure LearnIng of Probabilistic logic programs by searChing OVER the clause space): an algorithm for learning the structure of programs by searching the clause space and the theory space separatery \cite{BelRig13-TPLP-IJ}
\end{itemize} \end{itemize}
\subsection{Input} \subsection{Input}
@ -448,7 +449,7 @@ To execute the learning algorithms, prepare four files in the same folder:
\item \texttt{<stem>.kb}: contains the example interpretations \item \texttt{<stem>.kb}: contains the example interpretations
\item \texttt{<stem>.bg}: contains the background knowledge, i.e., knowledge valid for all interpretations \item \texttt{<stem>.bg}: contains the background knowledge, i.e., knowledge valid for all interpretations
\item \texttt{<stem>.l}: contains language bias information \item \texttt{<stem>.l}: contains language bias information
\item \texttt{<stem>.cpl}: contains the LPAD for you which you want to learn the parameters or the initial LPAD for SLIPCASE \item \texttt{<stem>.cpl}: contains the LPAD for you which you want to learn the parameters or the initial LPAD for SLIPCASE. For SLIPCOVER, this file should be absent
\end{itemize} \end{itemize}
where \texttt{<stem>} is your dataset name. Examples of these files can be found in the dataset pages. where \texttt{<stem>} is your dataset name. Examples of these files can be found in the dataset pages.
@ -503,7 +504,7 @@ For RIB, if there are unseen predicates, i.e., predicates that are present in th
unseen(<predicate>/<arity>). unseen(<predicate>/<arity>).
\end{verbatim} \end{verbatim}
For SLIPCASE, you have to specify the language bias by means of mode declarations in the style of For SLIPCASE and SLIPCOVER, you have to specify the language bias by means of mode declarations in the style of
\href{http://www.doc.ic.ac.uk/\string ~shm/progol.html}{Progol}. \href{http://www.doc.ic.ac.uk/\string ~shm/progol.html}{Progol}.
\begin{verbatim} \begin{verbatim}
modeh(<recall>,<predicate>(<arg1>,...). modeh(<recall>,<predicate>(<arg1>,...).
@ -529,6 +530,16 @@ for specifying an output variable of type \texttt{<type>}. or
\end{verbatim} \end{verbatim}
for specifying a constant. for specifying a constant.
SLIPCOVER also allows the arguments
\begin{verbatim}
#<type>
\end{verbatim}
for specifying an argument which should be replaced by a constant of type \texttt{<type>} in the bottom clause but should not be used for replacing input variables of the following literals or
\begin{verbatim}
-#<type>
\end{verbatim}
for specifying an argument which should be replaced by a constant of type \texttt{<type>} in the bottom clause and that should be used for replacing input variables of the following literals. \verb|#| and \verb|-#| differ only in the creation of the bottom clause.
An example of language bias for the UWCSE domain is An example of language bias for the UWCSE domain is
\begin{verbatim} \begin{verbatim}
output(advisedby/2). output(advisedby/2).
@ -547,6 +558,31 @@ modeb(*,samecourse(+course, -course)).
modeb(*,samecourse(-course, +course)). modeb(*,samecourse(-course, +course)).
.... ....
\end{verbatim} \end{verbatim}
SLIPCOVER also requires facts for the \verb|determination/2| predicate that indicate which predicates can appear in the body of clauses.
For example
\begin{verbatim}
determination(professor/1,student/1).
determination(student/1,hasposition/2).
\end{verbatim}
state that \verb|student/1| can appear in the body of clauses for \verb|professor/1| and that \verb|hasposition/2| can appear in
the body of clauses for \verb|student/1|.
SLIPCOVER also allows mode declarations of the form
\begin{verbatim}
modeh(<r>,[<s1>,...,<sn>],[<a1>,...,<an>],[<P1/Ar1>,...,<Pk/Ark>]).
\end{verbatim}
These mode declarations are used to generate clauses with more than two head atoms. In them, \verb|<s1>,...,<sn>| are schemas, \verb|<a1>,...,<an>| are atoms such that \verb|<ai>| is obtained from $\verb|<si>|$ by replacing placemarkers with variables,
\verb|<Pi/Ari>| are the predicates admitted in the body. \verb|<a1>,...,<an>| are used to indicate which variables should be shared by the atoms in the head.
An example of such a mode declaration is
\begin{verbatim}
modeh(*,
[advisedby(+person,+person),tempadvisedby(+person,+person)],
[advisedby(A,B),tempadvisedby(A,B)],
[professor/1,student/1,hasposition/2,inphase/2,
publication/2,taughtby/3,ta/3,courselevel/2,yearsinprogram/2]).
\end{verbatim}
\subsection{Parameters} \subsection{Parameters}
In order to set the algorithms' parameters, you have to insert in \texttt{<stem>.l} commands of the form In order to set the algorithms' parameters, you have to insert in \texttt{<stem>.l} commands of the form
@ -556,20 +592,17 @@ In order to set the algorithms' parameters, you have to insert in \texttt{<stem>
The available parameters are: The available parameters are:
\begin{itemize} \begin{itemize}
\item \verb|depth| (values: integer or \verb|inf|, default value: 3): depth of derivations if \verb|depth_bound| is set to \verb|true| \item \verb|depth| (values: integer or \verb|inf|, default value: 3): depth of derivations if \verb|depth_bound| is set to \verb|true|
\item \verb|single_var| (values: \verb|{true,false}|, default value: \verb|false|, valid for CEM, EMBLEM and SLIPCASE): if set to \verb|true|, there is a random variable for each clauses, instead of a separate random variable for each grounding of a clause \item \verb|single_var| (values: \verb|{true,false}|, default value: \verb|false|, valid for CEM, EMBLEM, SLIPCASE and SLIPCOVER): if set to \verb|true|, there is a random variable for each clauses, instead of a separate random variable for each grounding of a clause
\item \verb|sample_size| (values: integer, default value: 1000): total number of examples in case in which the models in the \verb|.kb| file contain a \verb|prob(P).| fact. In that case, one model corresponds to \verb|sample_size*P| examples \item \verb|sample_size| (values: integer, default value: 1000): total number of examples in case in which the models in the \verb|.kb| file contain a \verb|prob(P).| fact. In that case, one model corresponds to \verb|sample_size*P| examples
\item \verb|epsilon_em| (values: real, default value: 0.1, valid for CEM, EMBLEM and SLIPCASE): if the difference in the log likelihood in two successive EM iteration is smaller \item \verb|epsilon_em| (values: real, default value: 0.1, valid for CEM, EMBLEM, SLIPCASE and SLIPCOVER): if the difference in the log likelihood in two successive EM iteration is smaller
than \verb|epsilon_em|, then EM stops than \verb|epsilon_em|, then EM stops
\item \verb|epsilon_em_fraction| (values: real, default value: 0.01, valid for CEM, EMBLEM and SLIPCASE): if the difference in the log likelihood in two successive EM iteration is smaller \item \verb|epsilon_em_fraction| (values: real, default value: 0.01, valid for CEM, EMBLEM, SLIPCASE and SLIPCOVER): if the difference in the log likelihood in two successive EM iteration is smaller
than \verb|epsilon_em_fraction|*(-current log likelihood), then EM stops than \verb|epsilon_em_fraction|*(-current log likelihood), then EM stops
\item \verb|iter| (values: integer, defualt value: 1, valid for EMBLEM and SLIPCASE): maximum number of iteration of EM parameter learning. If set to -1, no maximum number of iterations is imposed \item \verb|iter| (values: integer, defualt value: 1, valid for EMBLEM, SLIPCASE and SLIPCOVER): maximum number of iteration of EM parameter learning. If set to -1, no maximum number of iterations is imposed
\item \verb|iterREF| (values: integer, defualt value: 1, valid for SLIPCASE): \item \verb|iterREF| (values: integer, defualt value: 1, valid for SLIPCASE and SLIPCOVER):
maximum number of iteration of EM parameter learning for refinements. If set to -1, no maximum number of iterations is imposed. maximum number of iteration of EM parameter learning for refinements. If set to -1, no maximum number of iterations is imposed.
\item \verb|random_restarts_number| (values: integer, default value: 1, valid for CEM, EMBLEM, SLIPCASE and SLIPCOVER): number of random restarts of EM learning
\item \verb|random_restarts_number| (values: integer, default value: 1, valid for CEM, EMBLEM and SLIPCASE): number of random restarts of EM learning \item \verb|random_restarts_REFnumber| (values: integer, default value: 1, valid for SLIPCASE and SLIPCOVER): number of random restarts of EM learning for refinements
\item \verb|random_restarts_REFnumber| (values: integer, default value: 1, valid for SLIPCASE): number of random restarts of EM learning for refinements
\item \verb|setrand| (values: rand(integer,integer,integer)): seed for the random functions, see Yap manual for allowed values \item \verb|setrand| (values: rand(integer,integer,integer)): seed for the random functions, see Yap manual for allowed values
\item \verb|minimal_step| (values: [0,1], default value: 0.005, valid for RIB): minimal increment of $\gamma$ \item \verb|minimal_step| (values: [0,1], default value: 0.005, valid for RIB): minimal increment of $\gamma$
\item \verb|maximal_step| (values: [0,1], default value: 0.1, valid for RIB): maximal increment of $\gamma$ \item \verb|maximal_step| (values: [0,1], default value: 0.1, valid for RIB): maximal increment of $\gamma$
@ -577,15 +610,23 @@ than \verb|epsilon_em_fraction|*(-current log likelihood), then EM stops
\item \verb|delta| (values: negative integer, default value -10, valid for RIB): value assigned to $\log 0$ \item \verb|delta| (values: negative integer, default value -10, valid for RIB): value assigned to $\log 0$
\item \verb|epsilon_fraction| (values: integer, default value 100, valid for RIB): in the computation of the step, the value of $\epsilon$ of \cite{DBLP:journals/jmlr/ElidanF05} is obtained as $\log |CH,T|\times$\verb|epsilon_fraction| \item \verb|epsilon_fraction| (values: integer, default value 100, valid for RIB): in the computation of the step, the value of $\epsilon$ of \cite{DBLP:journals/jmlr/ElidanF05} is obtained as $\log |CH,T|\times$\verb|epsilon_fraction|
\item \verb|max_rules| (values: integer, default value: 6000, valid for RIB and SLIPCASE): maximum number of ground rules. Used to set the size of arrays for storing internal statistics. Can be increased as much as memory allows. \item \verb|max_rules| (values: integer, default value: 6000, valid for RIB and SLIPCASE): maximum number of ground rules. Used to set the size of arrays for storing internal statistics. Can be increased as much as memory allows.
\item \verb|logzero| (values: negative real, default value $\log(0.000001)$, valid for SLIPCASE): value assigned to $\log 0$ \item \verb|logzero| (values: negative real, default value $\log(0.000001)$, valid for SLIPCASE and SLIPCOVER): value assigned to $\log 0$
\item \verb|examples| (values: \verb|atoms|,\verb|interpretations|, default value \verb|atoms|, valid for SLIPCASE): determines how BDDs are built: if set to \verb|interpretations|, a BDD for the conjunction of all the atoms for the target predicates in each interpretations is built. \item \verb|examples| (values: \verb|atoms|,\verb|interpretations|, default value \verb|atoms|, valid for SLIPCASE): determines how BDDs are built: if set to \verb|interpretations|, a BDD for the conjunction of all the atoms for the target predicates in each interpretations is built.
If set to \verb|atoms|, a BDD is built for the conjunction of a group of atoms for the target predicates in each interpretations. The number of atoms in each group is determined by the parameter \verb|group| If set to \verb|atoms|, a BDD is built for the conjunction of a group of atoms for the target predicates in each interpretations. The number of atoms in each group is determined by the parameter \verb|group|
\item \verb|group| (values: integer, default value: 1, valid for SLIPCASE): number of target atoms in the groups that are used to build BDDs \item \verb|group| (values: integer, default value: 1, valid for SLIPCASE): number of target atoms in the groups that are used to build BDDs
\item \verb|nax_iter| (values: integer, default value: 10, valid for SLIPCASE): number of interations of beam search \item \verb|nax_iter| (values: integer, default value: 10, valid for SLIPCASE and SLIPCOVER): number of interations of beam search
\item \verb|max_var| (values: integer, default value: 1, valid for SLIPCASE): maximum number of distinct variables in a clause \item \verb|max_var| (values: integer, default value: 1, valid for SLIPCASE and SLIPCOVER): maximum number of distinct variables in a clause
\item \verb|verbosity| (values: integer in [1,3], default value: 1): level of verbosity of the algorithms \item \verb|verbosity| (values: integer in [1,3], default value: 1): level of verbosity of the algorithms
\item \verb|beamsize| (values: integer, default value: 20, valid for SLIPCASE): size of the beam in SLIPCASE \item \verb|beamsize| (values: integer, default value: 20, valid for SLIPCASE and SLIPCOVER): size of the beam
\item \verb|megaex_bottom| (values: integer, default value: 1, valid for SLIPCOVER): number of mega-examples on which to build the bottom clauses
\item \verb|initial_clauses_per_megaex| (values: integer, default value: 1, valid for SLIPCOVER):
number of bottom clauses to build for each mega-example
\item \verb|d| (values: integer, default value: 10000, valid for SLIPCOVER):
number of saturation steps when building the bottom clause
\item \verb|max_iter_structure| (values: integer, default value: 1, valid for SLIPCOVER):
maximum number of theory search iterations
\item \verb|background_clauses| (values: integer, default value: 50, valid for SLIPCOVER):
maximum numbers of background clauses.
\end{itemize} \end{itemize}
\subsection{Commands} \subsection{Commands}
To execute CEM, load \texttt{em.pl} with To execute CEM, load \texttt{em.pl} with
@ -620,8 +661,16 @@ and call
\begin{verbatim} \begin{verbatim}
?:- sl(stem). ?:- sl(stem).
\end{verbatim} \end{verbatim}
To execute SLIPCOVER, load \texttt{slipcover.pl} with
\begin{verbatim}
?:- use_module(library('cplint/slipcover')).
\end{verbatim}
and call
\begin{verbatim}
?:- sl(stem).
\end{verbatim}
\subsection{Learning Examples} \subsection{Learning Examples}
The subfolders \verb|em|, \verb|rib| and \verb|slipcase| of the \verb|packages/cplint| folder in Yap git distribution The subfolders \verb|em|, \verb|rib|, \verb|slipcase| and \verb|slipcover| of the \verb|packages/cplint| folder in Yap git distribution
contain examples of input and output files for the learning algorithms. contain examples of input and output files for the learning algorithms.
\section{License} \section{License}