Merge pull request opencog#3259 from ngeiswei/improve-miner-utests

Improve miner utests

opencog/learning/miner/Miner.cc

@@ -48,8 +48,9 @@ namespace opencog
 // TODO:
 // 7. make sure that filtering is still meaningfull
 
-XPMParameters::XPMParameters(unsigned ms, unsigned iconjuncts,
-                             const Handle& ipat, int maxd, double io)
+MinerParameters::MinerParameters(unsigned ms, unsigned iconjuncts,
+                                 const Handle& ipat, int maxd,
+                                 double io)
 	: minsup(ms), initconjuncts(iconjuncts), initpat(ipat),
 	  maxdepth(maxd), info(io)
 {
@@ -79,14 +80,19 @@ XPMParameters::XPMParameters(unsigned ms, unsigned iconjuncts,
 	initconjuncts = Miner::conjuncts(initpat);
 }
 
-Miner::Miner(AtomSpace& as, const XPMParameters& prm)
-	: text_as(as), param(prm) {}
+Miner::Miner(const MinerParameters& prm)
+	: param(prm) {}
 
-HandleTree Miner::operator()()
+HandleTree Miner::operator()(const AtomSpace& texts_as)
 {
 	HandleSet texts;
-	text_as.get_handles_by_type(std::inserter(texts, texts.end()),
-	                            opencog::ATOM, true);
+	texts_as.get_handles_by_type(std::inserter(texts, texts.end()),
+	                             opencog::ATOM, true);
+	return operator()(texts);
+}
+
+HandleTree Miner::operator()(const HandleSet& texts)
+{
 	// If the initial pattern is specialized, this initial filtering
 	// may save some computation
 	HandleSet fltexts = filter_texts(param.initpat, texts);
@@ -97,6 +103,7 @@ HandleTree Miner::specialize(const Handle& pattern,
                              const HandleSet& texts,
                              int maxdepth)
 {
+	// TODO: decide what to choose and remove or comment
 	// return specialize_alt(pattern, texts, Valuations(pattern, texts), maxdepth);
 	return specialize(pattern, texts, Valuations(pattern, texts), maxdepth);
 }
@@ -651,25 +658,25 @@ Handle Miner::restricted_satisfying_set(const Handle& pattern,
                                         const HandleSet& texts,
                                         int maxf)
 {
-	static AtomSpace tmp_text_as;
-	tmp_text_as.clear();
+	static AtomSpace tmp_texts_as;
+	tmp_texts_as.clear();
 	HandleSeq tmp_texts;
 	for (const auto& text : texts)
-		tmp_texts.push_back(tmp_text_as.add_atom(text));
+		tmp_texts.push_back(tmp_texts_as.add_atom(text));
 
 	// Avoid pattern matcher warning
 	// TODO: support 1 < conjuncts
 	if (totally_abstract(pattern) and conjuncts(pattern) == 1)
-		return tmp_text_as.add_link(SET_LINK, tmp_texts);
+		return tmp_texts_as.add_link(SET_LINK, tmp_texts);
 
 	// Run the pattern matcher
-	AtomSpace tmp_query_as(&tmp_text_as);
+	AtomSpace tmp_query_as(&tmp_texts_as);
 	Handle tmp_pattern = tmp_query_as.add_atom(pattern),
 		vardecl = get_vardecl(tmp_pattern),
 		body = get_body(tmp_pattern),
 		gl = tmp_query_as.add_link(GET_LINK, vardecl, body),
-		results = (maxf < 0 ? satisfying_set(&tmp_text_as, gl)
-		           : satisfying_set(&tmp_text_as, gl, maxf));
+		results = (maxf < 0 ? satisfying_set(&tmp_texts_as, gl)
+		           : satisfying_set(&tmp_texts_as, gl, maxf));
 	return results;
 }
 

opencog/learning/miner/Miner.h

@@ -41,17 +41,17 @@ namespace opencog
  * Frequent Subtree Mining -- An Overview, from Yun Chi et al, when
  * possible.
  */
-struct XPMParameters {
+struct MinerParameters {
 
 	/**
 	 * CTor. Note that conjuncts will be overwritten by initpat, if
 	 * provided.
 	 */
-	XPMParameters(unsigned minsup=1,
-	              unsigned conjuncts=1,
-	              const Handle& initpat=Handle::UNDEFINED,
-	              int maxdepth=-1,
-	              double info=1.0);
+	MinerParameters(unsigned minsup=1,
+	                unsigned conjuncts=1,
+	                const Handle& initpat=Handle::UNDEFINED,
+	                int maxdepth=-1,
+	                double info=1.0);
 
 	// Minimum support. Mined patterns must have a frequency equal or
 	// above this value.
@@ -114,15 +114,20 @@ class Miner
 	/**
 	 * CTor
 	 */
-	Miner(AtomSpace& as, const XPMParameters& param=XPMParameters());
+	Miner(const MinerParameters& param=MinerParameters());
 
 	/**
-	 * Mine and return a tree of patterns linked by specialization
-	 * relationship (children are specializations of parent) with
-	 * frequency equal to or above minsup, starting from the initial
-	 * pattern, excluded.
+	 * Mine the given AtomSpace and return a tree of patterns linked by
+	 * specialization relationship (children are specializations of
+	 * parent) with frequency equal to or above minsup, starting from
+	 * the initial pattern, excluded.
 	 */
-	HandleTree operator()();
+	HandleTree operator()(const AtomSpace& texts_as);
+
+	/**
+	 * Like above but only mine amongst the provided text collection.
+	 */
+	HandleTree operator()(const HandleSet& texts);
 
 	/**
 	 * Specialization. Given a pattern and a collection to text atoms,
@@ -148,14 +153,8 @@ class Miner
 	                          const Valuations& valuations,
 	                          int maxdepth);
 
-	// AtomSpace containing the text trees to mine.
-	AtomSpace& text_as;
-
 	// Parameters
-	XPMParameters param;
-
-	// Working atomspace containing the patterns, and other junk.
-	AtomSpace pattern_as;
+	MinerParameters param;
 
 private:
 
@@ -216,8 +215,10 @@ class Miner
 	 * at the given variable, then call Miner::specialize on the
 	 * obtained specialization.
 	 */
-	HandleTree specialize_shapat(const Handle& pattern, const HandleSet texts,
-	                             const Handle& var, const Handle& shapat,
+	HandleTree specialize_shapat(const Handle& pattern,
+	                             const HandleSet texts,
+	                             const Handle& var,
+	                             const Handle& shapat,
 	                             int maxdepth);
 
 	/**

tests/learning/miner/CMakeLists.txt

@@ -2,8 +2,3 @@ ADD_CXXTEST(MinerUTest)
 TARGET_LINK_LIBRARIES(MinerUTest
   miner
 )
-
-ADD_CXXTEST(UREMinerUTest)
-TARGET_LINK_LIBRARIES(UREMinerUTest
-  miner
-)