TakeLab software and resources

This page contains a summary of freely available software and resources created by TakeLab at the University of Zagreb, Croatia.

Tools and libraries

• Coral - A corpus aligner. Currently offline
• CroNER - A Croatian Named Entity Recognizer.
• DiaCRO - A diacritics restaurator for Croatian. Currently offline
• GPKEX - Genetically programmed keyphrase extraction.
• libsentences - A library for sentence boundary detection.
• MINERAL - A tool for extracting disease mentions from clinical text.
• TakeLab STS - A semantic textual similarity system from SemEval 2012 shared task.
• TermeX - A terminology extraction tool.
• TweetingJay - A tool for recognizing semantically identical tweets (in English).

Resources

• agenda - Preprocessed news articles annotated with topics.
• argpremises - A corpus of matched claims with implicit premises.
• ComArg - A corpus of online user comments with arguments.
• Cro6WSD - A small Croatian word sense disambiguation dataset.
• Cro36WSD - A medium multi-label Croatian word sense disambiguation datasets.
• CroCoref - A corpus with manually annotated entity coreference.
• CroLexSub - A preliminary lexical substitution dataset for Croatian. New
• CroMWE - Human-annotated examples of MWEs and non-MWE n-grams in Croatian, extended with linguistic feature labels.New
• CroMWEsc - A dataset annotated with semantic compositionality of Croatian Multiword Expressions.
• Cropinion - Opinion mining from Croatian user reviews dataset.
• CroSentCmp - A comparison of different sentiment classification methods for Croatian. New
• CroSyn - A synonym choice dataset for Croatian.
• CroSemRel450 - A word semantic relatedness dataset for Croatian.
• CroWSI - Graph-based induction of word senses in Croatian.
• DerivBase.hr - A large-coverage derivational morphology resource for Croatian.
• dm.hr - A distributional memory for Croatian.
• Event-centered information retrieval evaluation collections - Two collections of queries and documents in English for event-centered information retrieval.
• An event coreference dataset - A dataset with annotated event coreference (English).
• A factual event anchor extraction dataset - A dataset of 750 English newswire texts annotated for factual event mentions and 105 manually annotated event graphs.
• FAQ retrieval dataset - A dataset with queries and relevance judgements for FAQ retrieval in Croatian.
• fHrWaC - A filtered version of the hrWaC corpus (Croatian).
• HeidelTime.hr - Croatian resources for the HeidelTime tagger.
• kex.hr - A keyphrase extraction evaluation dataset for Croatian.
• NN13205 - An indexed Croatian legislative collection.
• HOFM - A higher-order functional morphology framework.
• MOLEX - A morphological lexicon for Croatian. Currently offline
• Recognizing identical events dataset - A dataset with annotated identical and similar events (English).
• Semantic analogies dataset - A dataset with semantic analogies.
• Sentiment Lexica - A prior sentiment lexica for English and Croatian.
• VerbCROcean - A repository of fine-grained semantic verb relations for Croatian.
• WikiWarsHr - A temporally tagged corpus of historical narratives (Croatian).
• wmic-lexsub - Representing Word Meaning in Context via Lexical Substitutes New
• PreTox - A dataset for preemptive toxic language detection in wikipedia comments.

Last revision: 25 September 2019