This is an automated email from the ASF dual-hosted git repository.
sergeykamov pushed a commit to branch master-model
in repository https://gitbox.apache.org/repos/asf/incubator-nlpcraft.git
The following commit(s) were added to refs/heads/master-model by this push:
new d68d697 WIP.
d68d697 is described below
commit d68d697a1498b84da576fd137d87d3a0226613c9
Author: Sergey Kamov <[email protected]>
AuthorDate: Fri Oct 8 10:59:06 2021 +0300
WIP.
---
.../src/main/scala/org/apache/nlpcraft/model/NCModel.java | 14 +++++---------
.../org/apache/nlpcraft/model/builders/NCModelBuilder.java | 8 ++------
.../nlpcraft/model/impl/opennlp/NCOpenNlpTokensParser.java | 8 ++++----
.../model/nlp/{NCNlpToken.java => NCNlpNerToken.java} | 4 ++--
.../{NCNlpTokensParser.java => NCNlpNerTokensParser.java} | 7 +++++--
.../scala/org/apache/nlpcraft/model/nlp/NCNlpRichWord.java | 6 +++---
.../scala/org/apache/nlpcraft/model/nlp/NCNlpWord.java | 1 +
.../org/apache/nlpcraft/model/nlp/NCNlpWordsParser.java | 3 +++
8 files changed, 25 insertions(+), 26 deletions(-)
diff --git a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/NCModel.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/NCModel.java
index a2bc4a4..e46912a 100644
--- a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/NCModel.java
+++ b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/NCModel.java
@@ -17,7 +17,7 @@
package org.apache.nlpcraft.model;
-import org.apache.nlpcraft.model.nlp.NCNlpTokensParser;
+import org.apache.nlpcraft.model.nlp.NCNlpNerTokensParser;
import org.apache.nlpcraft.model.nlp.NCNlpWordsParser;
import org.apache.nlpcraft.model.nlp.NCNlpWord;
@@ -72,10 +72,9 @@ public interface NCModel {
int DFLT_MIN_NON_STOPWORDS = 0;
boolean DFLT_IS_SWEAR_WORDS_ALLOWED = false;
boolean DFLT_IS_NO_NOUNS_ALLOWED = true;
- boolean DFLT_IS_NO_USER_TOKENS_ALLOWED = true;
boolean DFLT_IS_STOPWORDS_ALLOWED = true;
- // Common.
+ // Common description.
String getId();
String getName();
String getVersion();
@@ -131,18 +130,16 @@ public interface NCModel {
return DFLT_MAX_TOTAL_SYNONYMS;
}
- default boolean isNoUserTokensAllowed() {
- return DFLT_IS_NO_USER_TOKENS_ALLOWED;
- }
-
default Map<String, Object> getMetadata() {
return DFLT_METADATA;
}
+ // TODO: drop it.
default Set<String> getAdditionalStopWords() {
return Collections.emptySet();
}
+ // TODO: drop it.
default Set<String> getExcludedStopWords() {
return Collections.emptySet();
}
@@ -168,12 +165,11 @@ public interface NCModel {
// Nlp.
NCNlpWordsParser getNlpWordsParser();
- List<NCNlpTokensParser> getNlpTokensParsers();
+ List<NCNlpNerTokensParser> getNlpTokensParsers();
Function<List<NCNlpWord>, List<NCNlpWord>> getStopWordsFilter();
Function<List<NCNlpWord>, List<NCNlpWord>> getSwearWordsFilter();
Function<List<NCNlpWord>, List<NCNlpWord>> getSuspiciousWordsFilter();
-
// Intents data.
// model class by default + for methods of given instances.
// Scanned for NCIntent, NCIntentRef, NCIntentSample, NCIntentSampleRef
diff --git
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/builders/NCModelBuilder.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/builders/NCModelBuilder.java
index 774747e..95daf89 100644
---
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/builders/NCModelBuilder.java
+++
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/builders/NCModelBuilder.java
@@ -1,7 +1,7 @@
package org.apache.nlpcraft.model.builders;
import org.apache.nlpcraft.model.NCContext;
-import org.apache.nlpcraft.model.nlp.NCNlpTokensParser;
+import org.apache.nlpcraft.model.nlp.NCNlpNerTokensParser;
import org.apache.nlpcraft.model.NCElement;
import org.apache.nlpcraft.model.NCIntentMatch;
import org.apache.nlpcraft.model.NCModel;
@@ -76,10 +76,6 @@ public class NCModelBuilder {
return null;
}
- public NCModelBuilder withNoUserTokensAllowed(boolean noUserTokensAllowed)
{
- return null;
- }
-
public NCModelBuilder withMetadata(Map<String, Object> meta) {
return null;
}
@@ -115,7 +111,7 @@ public class NCModelBuilder {
public NCModelBuilder withNlpWordsParser(NCNlpWordsParser parser) {
return null;
}
- public NCModelBuilder withNlpTokensParsers(List<NCNlpTokensParser>
parsers) {
+ public NCModelBuilder withNlpTokensParsers(List<NCNlpNerTokensParser>
parsers) {
return null;
}
public NCModelBuilder withStopWordsFilter(Function<List<NCNlpWord>,
List<NCNlpWord>> filter) {
diff --git
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/impl/opennlp/NCOpenNlpTokensParser.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/impl/opennlp/NCOpenNlpTokensParser.java
index d3b87a9..97340ca 100644
---
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/impl/opennlp/NCOpenNlpTokensParser.java
+++
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/impl/opennlp/NCOpenNlpTokensParser.java
@@ -3,14 +3,14 @@ package org.apache.nlpcraft.model.impl.opennlp;
import org.apache.nlpcraft.model.NCModel;
import org.apache.nlpcraft.model.NCRequest;
import org.apache.nlpcraft.model.nlp.NCNlpRichWord;
-import org.apache.nlpcraft.model.nlp.NCNlpToken;
-import org.apache.nlpcraft.model.nlp.NCNlpTokensParser;
+import org.apache.nlpcraft.model.nlp.NCNlpNerToken;
+import org.apache.nlpcraft.model.nlp.NCNlpNerTokensParser;
import java.util.List;
-public class NCOpenNlpTokensParser implements NCNlpTokensParser {
+public class NCOpenNlpTokensParser implements NCNlpNerTokensParser {
@Override
- public List<NCNlpToken> parse(NCRequest req, NCModel mdl,
List<NCNlpRichWord> words, List<NCNlpToken> elements) {
+ public List<NCNlpNerToken> parse(NCRequest req, NCModel mdl,
List<NCNlpRichWord> words, List<NCNlpNerToken> elements) {
return null;
}
}
diff --git
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpToken.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpNerToken.java
similarity index 92%
rename from
nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpToken.java
rename to
nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpNerToken.java
index bb02494..68cf4a5 100644
--- a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpToken.java
+++ b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpNerToken.java
@@ -21,8 +21,8 @@ import org.apache.nlpcraft.model.NCMetadata;
import java.util.*;
-public interface NCNlpToken extends NCMetadata {
- String getElementId();
+public interface NCNlpNerToken extends NCMetadata {
+ String getId();
List<NCNlpRichWord> getWords();
diff --git
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpTokensParser.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpNerTokensParser.java
similarity index 79%
rename from
nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpTokensParser.java
rename to
nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpNerTokensParser.java
index 334ccd2..0b0161c 100644
---
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpTokensParser.java
+++
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpNerTokensParser.java
@@ -22,6 +22,9 @@ import org.apache.nlpcraft.model.NCRequest;
import java.util.*;
-public interface NCNlpTokensParser {
- List<NCNlpToken> parse(NCRequest req, NCModel mdl, List<NCNlpRichWord>
words, List<NCNlpToken> elements);
+/**
+ * User implementations and built in Stanford and OpneNlp implementations.
+ */
+public interface NCNlpNerTokensParser {
+ List<NCNlpNerToken> parse(NCRequest req, NCModel mdl, List<NCNlpRichWord>
words, List<NCNlpNerToken> elements);
}
diff --git
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpRichWord.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpRichWord.java
index fa4c0c6..0fba051 100644
--- a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpRichWord.java
+++ b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpRichWord.java
@@ -19,6 +19,9 @@ package org.apache.nlpcraft.model.nlp;
import org.apache.nlpcraft.model.nlp.NCNlpWord;
+/**
+ * Extended word data.
+ */
public interface NCNlpRichWord extends NCNlpWord {
boolean isStopWord();
@@ -29,7 +32,4 @@ public interface NCNlpRichWord extends NCNlpWord {
boolean isKnownWord();
boolean isSwearWord();
-
- // TODO:
- // boolean isEnglish();
}
diff --git
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWord.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWord.java
index 73418ac..77fdf75 100644
--- a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWord.java
+++ b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWord.java
@@ -1,5 +1,6 @@
package org.apache.nlpcraft.model.nlp;
+// Initial parser result by NCNlpWordsParser.
public interface NCNlpWord {
String getWord();
String getNormalWord();
diff --git
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWordsParser.java
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWordsParser.java
index fc283c0..d33cf52 100644
---
a/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWordsParser.java
+++
b/nlpcraft/src/main/scala/org/apache/nlpcraft/model/nlp/NCNlpWordsParser.java
@@ -4,6 +4,9 @@ import org.apache.nlpcraft.model.NCRequest;
import java.util.List;
+/**
+ * Initial request parser.
+ */
public interface NCNlpWordsParser {
List<NCNlpWord> parse(NCRequest req);
}
