How to Programming with Stanford-nlp

Try parameterizing the PTBTokenizer class. For example: PTBTokenizer<CoreLabel> ptbt = new PTBTokenizer<>(new FileReader(arg), new CoreLabelTokenFactory(), ""); ...

configuring a separate model jar in stanford nlp

nlp,stanford-nlp,sentiment-analysis,pos-tagger

See http://nlp.stanford.edu/software/corenlp.shtml#caseless Copying from the documentation: It is possible to run StanfordCoreNLP with tagger, parser, and NER models that ignore capitalization. In order to do this, download the caseless models package. Be sure to include the path to the case insensitive models jar in the -cp classpath flag as well....

NLP- Sentiment Processing for Junk Data takes time

Yes, the standard PCFG parser (the one that is run by default without any other options specified) will choke on this sort of long nonsense data. You might have better luck using the shift-reduce constituency parser, which is substantially faster than the PCFG and nearly as accurate.

Swapping in Berkley parser in Stanford corenlp

The difficult but clean way to do this would be to build your own annotator which hooks into a programmatic API of the Berkeley parser. You'd basically want to imitate the behavior of the ParserAnnotator, replacing the references to the Stanford ParserQuery implementation with references to Berkeley Parser + code...

Stanford NLP: Sentence splitting without tokenization?

character-encoding,command-line-interface,stanford-nlp

Our pipeline requires that you tokenize first; we use these tokens in the sentence-splitting algorithm. If your text is pre-tokenized, you can use DocumentPreproccesor and request whitespace-only tokenization. Let me know if I misunderstood your question....

Why POS tagging algorithm tags `can't` as separate words?

stanford-nlp,pos-tagger

Note: This is not the perfect answer. I think that the problem originates from the Tokenizer used in Stanford POS Tagger, not from the tagger itself. the Tokenizer (PTBTokenizer) can not handle apostrophe properly: 1- Stanford PTBTokenizer token's split delimiter. 2- Stanford coreNLP - split words ignoring apostrophe. As they...

swiftly generate and sort full encoding dictionary and corresponding primary radicals

This information is encoded in exactly the form you want in the RadicalMap source code. See the static initializer: String[] radLists = {"\u4e00\u4e00\u4e01\u4e02\u4e03...", "...", ..., }; Each string in this list has as its first character a radical, and the remaining characters have that first character as their primary radical....

Choosing correct word for the given string

java-7,stanford-nlp,eclipse-3.4,lemmatization

You cannot really check for every word because there are certain words which have more than 1 alphabets in their spelling. So one way you could go is - check for each alphabet in the word and restrict its number of consecutive appearances to two now check the new spelling...

Stanford CorpNLP returning wrong results

The problem in this example is that the word painting can be the present participle of to paint or a noun and the output of the lemmatizer depends on the part-of-speech tag assigned to the original word. If you run the tagger only on the fragment painting, then there is...

How configure Stanford QNMinimizer to get similar results as scipy.optimize.minimize L-BFGS-B

java,optimization,machine-learning,scipy,stanford-nlp

What you have should be just fine. (Have you actually had any problems with it?) Setting termination both on max iterations and max function evaluations is probably overkill, so you might omit the last argument to qn.minimize(), but it seems from the documentation that scipy does use both with a...

Get list of annotators in Stanford CoreNLP

Looking at the code behind the pipeline, it looks it's not currently possible to get the list of annotators enabled for an already-constructed pipeline (i). All of the relevant members storing this information are private. You could probably hack something up to get the annotator dependencies (ii), but it wouldn't...

software to extract word functions like subject, predicate, object etc

According to http://nlp.stanford.edu/software/lex-parser.shtml, Stanford NLP does have a parser which can identify the subject and predicate of a sentence. You can try it out online http://nlp.stanford.edu:8080/parser/index.jsp. You can use the typed dependencies to identify the subject, predicate, and object. From the example page, the sentence My dog also likes eating...

Detect relation between two persons in text

nlp,stanford-nlp,opennlp

You can see if there's a dependency path between the two entities in the sentence. For more info: http://nlp.stanford.edu/software/stanford-dependencies.shtml It won't be 100% accurate but good enough. To improve accuracy, you can prune the paths that are longer than certain lengths or have certain dependencies. You can also look at...

java.lang.NullPointerException while doing sentimental analysis with standford-nlp API

nlp,stanford-nlp,sentiment-analysis,shift-reduce

What is the code that produces this output? My strong suspicion is that you have not included the "sentiment" annotator in your annotators list, either in the properties file you are using to run the code, or the properties object you have passed into the annotation pipeline. Without running the...

NLP Shift reduce parser is throwing null pointer Exception for Sentiment calculation

Is there a specific reason why you are using version 3.4.1 and not the latest version? If I run your code with the latest version, it works for me (after I change the path to the SR model to edu/stanford/nlp/models/srparser/englishSR.ser.gz but I assume you changed that path on purpose). Also...

Sentence-level to document-level sentiment analysis. Analysing news

stanford-nlp,sentiment-analysis

Update: You might want to look into http://blog.getprismatic.com/deeper-content-analysis-with-aspects/ This is a very active area of research so it would be hard to find an off-the-shelf tool to do this (at least nothing is built in the Stanford CoreNLP). Some pointers: look into aspect-based sentiment analysis. In this case, Apple would...

Instruction for training model in Stanford Core NLP

stanford-nlp,sentiment-analysis,training-data

The more complicated options are described in comments in these classes: RNNOptions, RNNTrainOptions. The remainder of the options you listed are paths for reading / writing during training. The -trainPath argument points to a labeled sentiment treebank. The trees in this data will be used to train the model parameters...

Different results performing Part of Speech tagging using Core NLP and Stanford Parser?

stanford-nlp,part-of-speech

This isn't really about CoreNLP, it's about whether you are using the Stanford POS tagger or the Stanford Parser (the PCFG parser) to do the POS tagging. (The PCFG parser usually does POS tagging as part of its parsing algorithm, although it can also use POS tags given from elsewhere.)...

Annotator for Relationship Extraction

regex,nlp,nltk,stanford-nlp,gate

There is no PR in GATE that that will pair arguments and create instances for you. You must therefore create instances that are relevant to your problem. You can: write a custom PR or write some JAPE with Java RHS You can probably split your corpus on a training and...

StanfordCoreNLP : TokenMgrError: Lexical error at line 1, column 14. Encountered: “E” (69), after : “\\”

I also replied on the github page: We are releasing the new version of the software in a few days. This bug is most probably fixed in that -- I used the files you provided in the github page with the new code and it works. Stay tuned!

How to get dependency parse output exactly as online demo?

java,nlp,stanford-nlp,stemming,lemmatization

The reason for the different output is that if you use the parser demo, the stand-alone parser distribution is being used and your code uses the entire CoreNLP distribution. While both of them use the same parser and the same models, the default configuration of CoreNLP runs a part-of-speech (POS)...

StanfordNLP lemmatization cannot handle -ing words

Lemmatization crucially depends on the part of speech of the token. Only tokens with the same part of speech are mapped to the same lemma. In the sentence "This is confusing", confusing is analyzed as an adjective, and therefore it is lemmatized to confusing. In the sentence "I was confusing...

Chinese sentence segmenter with Stanford coreNLP

java,nlp,tokenize,stanford-nlp

Using Stanford Segmenter instead: $ wget http://nlp.stanford.edu/software/stanford-segmenter-2015-04-20.zip $ unzip stanford-segmenter-2015-04-20.zip $ echo "应有尽有的丰富选择定将为您的旅程增添无数的赏心乐事" > input.txt $ bash stanford-segmenter-2015-04-20/segment.sh ctb input.txt UTF-8 0 > output.txt $ cat output.txt 应有尽有的丰富选择定将为您的旅程增添无数的赏心乐事 Other than Stanford Segmenter, there are many other...

Unknown symbol in nltk pos tagging for Arabic

python,nlp,nltk,stanford-nlp,pos-tagger

The default NLTK POS tag is trained on English texts and is supposedly for English text processing, see http://www.nltk.org/_modules/nltk/tag.html. The docs: An off-the-shelf tagger is available. It uses the Penn Treebank tagset: >>> from nltk.tag import pos_tag # doctest: +SKIP >>> from nltk.tokenize import word_tokenize # doctest: +SKIP >>> pos_tag(word_tokenize("John's...

Need a good relation extractor

nlp,nltk,stanford-nlp

There are a few different tools you might want to look at: MITIE MIT's new MITIE tool supports basic relationship extraction. Included in the distribution are 21 English binary relation extraction models trained on a combination of Wikipedia and Freebase data. You can also train your own custom relation detectors....

StanfordCoreNLP: Why multiple roots for SemanticGraph (e.g. dependency parsing)

In all honesty, SemanticGraph has a lot of historical code which was motivated by its initial use in an RTE (Recognizing Textual Entailment) system, not for syntactic dependency parsing, so don't read too much into it all. But, nevertheless, there are various fairly natural use cases (e.g., fragment parsing or...

Identify prepositons and individual POS

I have had some breakthrough to understand if the word is actually preposition or subordinating conjunction. I have parsed following sentence : She left early because Mike arrived with his new girlfriend. (here because is subordinating conjunction ) After POS tagging She_PRP left_VBD early_RB because_IN Mike_NNP arrived_VBD with_IN his_PRP$ new_JJ...

Stanford coreNLP : how to get Label, position, and typed dependecies from parse Tree

python,nlp,nltk,stanford-nlp,context-free-grammar

You can get the position of a CoreLabel within its containing sentence with the CoreAnnotations.IndexAnnotation annotation. Your method for finding all dependents of a given word seems correct, and is probably the easiest way to do it....

how to extract elements from tree.productions()

nltk.Tree is actually a subclass of the Python list, so you can access the children of any node c by c[0], c[1], c[2], etc. Note that NLTK trees are not explicitly binary by design, so your notion of "left" and "right" might have to be enforced somewhere in a contract....

Coreference resolution using Stanford CoreNLP

There is a Annotation constructor with a List<CoreMap> sentences argument which sets up the document if you have a list of already tokenized sentences. For each sentence you want to create a CoreMap object as following. (Note that I also added a sentence and token index to each sentence and...

What is the default behavior of Stanford NLP's WordsToSentencesAnnotator when splitting a text into sentences?

It does split on these characters, however only when they appear as their own token and not at the end of an abbreviation such as in "etc.". So the issue here is not the sentence splitter but the tokenizer which thinks that "N." is an abbreviation and therefore does not...

Converting Stanford dependency relation to dot format

parsing,stanford-nlp

You need to call toDotFormat on an entire dependency tree. How have you generated these dependency trees in the first place? If you're using the StanfordCoreNLP pipeline, adding in the toDotFormat call is easy: Properties properties = new Properties(); props.put("annotators", "tokenize, ssplit, pos, depparse"); StanfordCoreNLP pipeline = new StanfordCoreNLP(props); String...

How to Identify mentions in a text?

twitter,machine-learning,stanford-nlp,sentiment-analysis

I think you can get what you want from the standard dcoref annotator. Look at the annotation set by this annotator, CorefChainAnnotation. This is a map from document entities to "coref chains." Each CorefChain can provide you with a list of mentions for the relevant entity in textual order....

NLP Postagger can't grok imperatives?

stanford-nlp,pos-tagger

There is no special tag for imperatives, they are simply tagged as VB. The info on the website refers to the fact that we added a bunch of manually annotated imperative sentences to our training data such that the POS tagger gets more of them right, i.e. tags the verb...

Typed Dependency Parsing in NLTK Python

python,nltk,stanford-nlp

There exists a python wrapper for the Stanford parser, you can get it here. It will give you the dependency tree of your sentence. EDIT: I assume here that you launched a server as said here. I also assume that you have installed jsonrpclib. The following code will produce what...

How to integrate the GATE Twitter PoS model with Stanford NLP?

You can use GATE twitter pos model with stanford package ./corenlp.sh -file tweets.txt -pos.model gate-EN-twitter.model -ssplit.newlineIsSentenceBreak always use v3.3.1 for GATE...

Stanford NLP: Tokenize output on a single line?

You can use DocumentPreprocessor, either programmatically or from the command line. From the CLI: $ echo "This is a test. And some more." | java edu.stanford.nlp.process.DocumentPreprocessor 2>/dev/null This is a test . And some more . You can do the same thing programmatically; see this SO answer....

NullPointerException with Stanford NLP Spanish POS tagging

stanford-nlp,sentiment-analysis

Yes, it seems like there's a bug in the 3.4.1 Spanish models. The Spanish 3.5.0 models actually seem to be compatible with Java 7. You can download the models used in 3.5 (stanford-spanish-corenlp-2014-10-23-models.jar) and put that on your classpath instead. This fixed the problem for me running Java 7 locally....

Sentiment Analysis in Spanish with Stanford coreNLP

Unfortunately there is no Stanford sentiment model available for Spanish. At the moment all the Spanish words are likely being treated as generic "unknown words" by the sentiment analysis algorithm, which is why you're seeing consistently bad performance. You can certainly train your own model (documented elsewhere on the Internet,...

Stanford Word Segmenter for Chinese in Python how to return results without punctuation

python,stanford-nlp,punctuation,chinese-locale

I think you'd be better off just removing the punctuation after the text has been segmented; I'm fairly sure the Stanford segmenter takes cues from punctuation in doing its job, so you wouldn't want to do so beforehand. The following works for me on UTF-8 text. For Chinese punctuation, use...

using Wordnet with standford nlp

java,stanford-nlp,wordnet

This does not seem to be set by any annotator in the code currently. A code search suggests that it was used in the NER system at one point, but is no longer set by anything.

Setting intercept in Stanford-NLP LogisticClassifier

stanford-nlp,logistic-regression

Yes; you can simply define a new feature (e.g., "bias" or "intercept"), and set the weight of that to be the intercept value from scikit-learn.

Stanford CoreNLP - the egw4-reut.512.clusters cannot be found

It turns out that the problem was caused by Maven. My code was located in a utility library which was wrapping over the Stanford CoreNLP to provide additional processing, and was working perfectly well by itself. However when adding this project as a dependency to my master project, Maven defaulted...

ssplit.eolonly with Chinese text

java,performance,parsing,stanford-nlp,sentiment-analysis

Unfortunately, not at the moment (Apr 2015). The current segmenter doesn't support preserving line information. This would be a good thing to fix at some point....

Lazy parsing with Stanford CoreNLP to get sentiment only of specific sentences

If you're looking to speed up constituency parsing, the single best improvement is to use the new shift-reduce constituency parser. It is orders of magnitude faster than the default PCFG parser. Answers to your later questions: Why is CoreNLP parsing not lazy? This is certainly possible, but not something that...

StanfordNLP does not extract relations between entities

parsing,nlp,stanford-nlp,sentiment-analysis,text-analysis

The RelationMentionsAnnotation is a sentence-level annotation. You should first iterate over the sentences in the Annotation object and then try to retrieve the annotation. Here's a basic example of how to iterate over sentences: // these are all the sentences in this document // a CoreMap is essentially a Map...

Stanford Parser - Factored model and PCFG

This FAQ answer explains the difference in a long paragraph. Relevant parts are quoted below: Can you explain the different parsers? This answer is specific to English. It mostly applies to other languages although some components are missing in some languages. The file englishPCFG.ser.gz comprises just an unlexicalized PCFG grammar....

why Standford NLP Parser gives different result(sentiment) for same statement used in kaggle Movie review

python,nlp,scikit-learn,classification,stanford-nlp

I assume you are using Bag of Words and the comma and the dot are one of your features (a column in your X matrix). +-------------------------+-----------+-----------+----+ | Document/Features | Genuinely | unnerving | . | +-------------------------+-----------+-----------+----+ | Genuinely unnerving . | 1 | 1 | 1 | | Genuinely unnerving...

How can the NamedEntityTag be used as EntityMention in RelationMention in the RelationExtractor?

tokenize,stanford-nlp,misspelling

You can match the text from the EntityMention with the NamedEntityText value from the NamedEntityTag.

StanfordNLP Tokenizer

I assume you're referring to the .flex file for the tokenizer? You need to generate new Java code from this specification before building again. Use the flexeverything Ant build task (see our build spec). You may also find Twokenize useful. This is a self-contained tokenizer for tweets. It's part of...

CoreNLP ConLL format with CollapsedCCProcessedDependenciesAnnotation

parsing,stanford-nlp

You can retrieve the CC-processed dependencies programmatically. This question should serve as a good example (see the code in the example using the CollapsedCCProcessedDependenciesAnnotation). Gabor's answer from the mailing list explains this behavior very well (i.e., why you can't output collapsed dependencies directly): Note that in general the collapsed cc...

Text tokenization with Stanford NLP : Filter unrequired words and characters

java,machine-learning,tokenize,stanford-nlp

This is a very domain-specific task that we don't perform for you in CoreNLP. You should be able to make this work with a regular expression filter and a stopword filter on top of the CoreNLP tokenizer. Here's an example list of English stopwords....

Does the Stanford NER CRF implementation use sentences in the training phase?

python,nlp,nltk,stanford-nlp,named-entity-recognition

Two newlines are considered boundary of an example. Your examples can be anything from phrases to the whole documents. So for your example, if you want two sentences as two examples: James PERSON lives O in O Chicago LOCATION . O Coffee O in O Trieste LOCATION is O great...

Chunking Stanford Named Entity Recognizer (NER) outputs from NLTK format

It looks long but it does the work: ner_output = [(u'Remaking', u'O'), (u'The', u'O'), (u'Republican', u'ORGANIZATION'), (u'Party', u'ORGANIZATION')] chunked, pos = [], "" for i, word_pos in enumerate(ner_output): word, pos = word_pos if pos in ['PERSON', 'ORGANIZATION', 'LOCATION'] and pos == prev_tag: chunked[-1]+=word_pos else: chunked.append(word_pos) prev_tag = pos clean_chunked =...

How to replace a word by its most representative mention using Stanford CoreNLP Coreferences module

The challenge is you need to make sure that the token isn't part of its representative mention. For example, the token "Judy" has "Judy 's" as its representative mention, so if you replace it in the phrase "Judy 's", you'll end up with the double "'s". You can check if...

Stanford CoreNLP wrong coreference resolution

Like many components in AI, the Stanford coreference system is only correct to a certain accuracy. In the case of coreference this accuracy is actually relatively low (~60 on standard benchmarks in a 0-100 range). To illustrate the difficulty of the problem, consider the following apparently similar sentence with a...

StanfordNLP Spanish Tokenizer

tokenize,stanford-nlp

I suppose you are working with the SpanishTokenizer rather than PTBTokenizer. SpanishTokenizer is heavily based on the FrenchTokenizer, which comes also from the PTBTokenizer (English). I've run all three with your sentence and seems that the PTBTokenizer give you the results you need, but not the others. As all of...

How to use serialized CRFClassifier with StanfordCoreNLP prop 'ner'

The property name is ner.model, not ner.models, so your code is still trying to load the default models. Let me know if this is documented incorrectly somewhere....

Separately tokenizing and pos-tagging with CoreNLP

java,ubuntu,stanford-nlp,opennlp,lingpipe

It seems to me that you would be better off separating the tokenization phase from your other downstream tasks (so I'm basically answering Question 2). You have two options: Tokenize using the Stanford tokenizer (example from Stanford CoreNLP usage page). The annotators options should only take 'tokenizer' in your case....

NLP libraries installation guidelines for java

I used OpenNLP in my project. I think this instructions will help you to go through OpenNLP Library. Follow this document Download OpenNLP Library and add it to your build path Download trained models and put it to a folder modelIn = new FileInputStream("path"); InputStream modelIn = null; try {...

How to extract an unlabelled/untyped dependency tree from a TreeAnnotation using Stanford CoreNLP?

There is no support for Spanish dependency parsing in CoreNLP at the moment. This includes typed dependency conversion from constituency parses. There is a head finder implemented (but not fully tested). You could hack an untyped dependency converter using this head finder, but we have no guarantees that this will...

Getting locations in standford core nlp

From what I see here, you should try out models which ignore the capitalization of words. You just need to add this models jar file to the existing one: caseless models. For future reference: the jar link may be broken, but the first link goes to the page, where a...

Stanford coreNLP : can a word in a sentence be part of multiple Coreference chains

machine-learning,nltk,stanford-nlp,text-classification,naivebayes

A word can be part of multiple coreference mentions. Consider for example the mention "the new acquisition by Microsoft". In this case, there are two candidates for mentions: the new acquisition by Microsoft and Microsoft. From this example it also follows that a word can be part of multiple coreference...

How to train a naive bayes classifier with pos-tag sequence as a feature?

If you know how to train and predict texts (or sentences in your case) using nltk's naive bayes classifier and words as features, than you can easily extend this approach in order to classify texts by pos-tags. This is because the classifier don't care about whether your feature-strings are words...

Error when using StanfordCoreNLP

nlp,stanford-nlp,named-entity-recognition

The tagger file has to be placed into your project root. project -- src --> SentimentAnaTest -- english-left3words/english-left3words-distsim.tagger Tested in Eclipse project....

How to suppress unmatched words in Stanford NER classifiers?

Hi I'll try to help out! So it sounds to me like you have a list of strings that should be called "CURRENCY", and you have a list of strings that should be called "COUNTRY", etc... And you want something to tag strings based off of your list. So when...

Stanford Core NLP example code SemanticGraph exception

No, the sentence is fine and this is unfortunately a bug in our dependency converter. The part-of-speech tagger outputs a really weird POS sequence causing the parser to produce a completely wrong parse tree which leads to this exception in the constituency-to-dependency converter. I fixed the bug in the converter...

In the CoreNLP pipeline, is it possible to use the Coref tool (dcoref) with the new dependency parser tool (depparse)? [closed]

Good question! This currently isn't possible in the pipeline, though it really ought to be. I'll bring it up in our next development meeting. For now, if you know that your pipeline doesn't require constituency parses, you can easily get around this by setting a property in the pipeline flags:...

Ignore words for lemmatizer

I think I found the solution with my friend's help. for(CoreMap sentence: sentences) { // Iterate over all tokens in a sentence for (CoreLabel token: sentence.get(TokensAnnotation.class)) { System.out.print(token.get(OriginalTextAnnotation.class) + "\t"); System.out.println(token.get(LemmaAnnotation.class)); } } You can get original form of the word by calling token.get(OriginalTextAnnotation.class)....

In CoreNLP what is the different between the default generated dependency trees?

python,nlp,stanford-nlp,pos-tagger,part-of-speech

See section 4 of the Stanford Dependencies manual: Different styles of dependency representation. The first three subsections map to basic, collapsed, and CC-processed dependency representations, respectively.

Stanford NLP: Chinese Part of Speech labels?

We use the tag set of the (Penn/LDC/Brandeis/UC Boulder) Chinese Treebank. See here for details on the tag set: http://www.cis.upenn.edu/~chinese/ This was documented in the parser FAQ, but I'll add it to the tagger FAQ....

Data format for Stanford POS-tagger

stanford-nlp,dataformat

You should have one sentence per line (your second example). Using the first format will certainly affect tagging results: you'll effectively build a unigram tagger, in which all tagging is done without any sentence context at all....

python corenlp batch parse

python,batch-processing,stanford-nlp

It was my mistake. I missed "raw_output" in parameter passing of batch_parse. So, it should be like this: for value in batch_parse(raw_text_directory, corenlp_dir,raw_output=True): print value ...

How should I figure out the POS tag of “last” in this sentence?

The short answer is "no." CoreNLP provides part of speech tags at a certain high but not perfect accuracy, and it will occasionally make mistakes. Beyond tweaking the tags yourself, there's no easy automatic way to have its accuracy go up. The longer answer is that you can always re-train...

How can I use Stanford NLP commercially?

You can either use the software under the GPL license, or you can purchase a commercial license. For the latter, you can contact us at the support email address found here.

StanfordCoreNLP: why two different data structures for cons. parse and dependency parse?

python,subprocess,stanford-nlp,python-multithreading

The dependency parses, when collapsed, are not necessarily DAGs. From the Stanford Dependencies manual: The collapsed and CCprocessed dependencies are not a DAG. The graphs can contain small cycles between two nodes (only). These don’t seem eliminable given the current representational choices. They occur with relative clauses such as the...

How to collect output from a Python subprocess

Add th.join() at the end otherwise you may kill the thread prematurely before it has processed all the output when the main thread exits: daemon threads do not survive the main thread (or remove th.setDaemon(True) instead of th.join()).

NLP - Error while Tokenization and Tagging etc [duplicate]

You're trying to use Stanford NLP tools version 3.5 or later using a version of Java 7 or earlier. Either upgrade to Java 8 or downgrade to Stanford tools version 3.4.1.

Stanford Parser - train input specification

chinese/train.conll, chinese/dev.conll: These are training/dev files in CoNLL 2006 format, as discussed in section 4.1 of the paper: http://cs.stanford.edu/~danqi/papers/emnlp2014.pdf . (In general we don't have permission to distribute data sets to others.) chinese/embeddings.txt: These are word embeddings trained with word2vec as described in section 3.2 of the same paper....

how to make a light-weighted stanford-nlp.jar