cogcomp-nlp
cogcomp-nlp copied to clipboard
CogComp
CogComp's Natural Language Processing Libraries and Demos: Modules include lemmatizer, ner, pos, prep-srl, quantifier, question type, relation-extraction, similarity, temporal normalizer, tokenizer, t...
For the pipeline server have a way to retrieve the version number of the pipeline. Bonus: and show this version number in the demo.
``` [error] Exception in thread "main" java.lang.IndexOutOfBoundsException: Index: 0, Size: 0 [error] at java.util.ArrayList.rangeCheck(ArrayList.java:653) [error] at java.util.ArrayList.get(ArrayList.java:429) [error] at edu.illinois.cs.cogcomp.edison.features.factory.DependencyPath.getFeatures(DependencyPath.java:73) [error] at edu.illinois.cs.cogcomp.edison.features.factory.DependencyPath.getFeatures(DependencyPath.java:51) [error] at edu.illinois.cs.cogcomp.edison.features.factory.FeatureExtractorPairConjunction.getCombinedFeaturesImplementation(FeatureExtractorPairConjunction.java:43) [error] at edu.illinois.cs.cogcomp.edison.features.factory.PairExtractor.getCombinedFeatures(PairExtractor.java:79) [error]...
Throws exceptions when loading it . . .
Seems to be faster than MapDB: http://jetbrains.github.io/xodus/
https://github.com/CogComp/cogcomp-nlp/blob/master/dataless-classifier/src/main/resources/hierarchies/20newsgroups/parentChildIdMap.txt https://github.com/CogComp/cogcomp-nlp/blob/master/dataless-classifier/data/hierarchies/20newsgroups/parentChildIdMap.txt
For example: "sicily" "Balearic" with Word2Vec gives 1.0.
Currently, uploading files on the Minio server through the DataStore wrapper seem to have an upper limit of 2GB. Moreover, the wrapper just arbitrary truncates the file at 2GB instead...
I am thinking about extending [FeatureExtractor](https://github.com/CogComp/cogcomp-nlp/blob/master/edison/src/main/java/edu/illinois/cs/cogcomp/edison/features/FeatureExtractor.java) so that it can be applied in other formats in addition to Constituents. In particular I want to have feaure extractors for - Relation...
I'd like corpus readers to have a method to inform the user of the views they populate. One way to do this is to specify a [CorpusReaderConfigurator](https://github.com/CogComp/cogcomp-nlp/blob/master/corpusreaders/src/main/java/edu/illinois/cs/cogcomp/nlp/corpusreaders/CorpusReaderConfigurator.java) property/field name and...
