Patch extractor #186
Open
Patch extractor #186
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Since MathMLTools is published on mvn central, it is easier to patch minor problems directly within the submodule. Note that this submodule is *not* on the master branch but on mathosphere-fix.
The <math> tags (as well as all other xml tags in <text>) are by default escaped when downloading a wiki dump. Hence it is not <math> but <math>. Since the AstVisitor is configured to find <math> (as well as <chem>) tags but <math> is not an xml tag, the visitor was unable to find them.
…t from 2.2.0 (why ever used in travis) to 3.3.0
… not only in the list of relations (those are two different lists)
Sweble is unable to detect 'Test' as the name of the reference in '<ref name=Test>'. Instead it interprets name and Test as attribute keys both with null-values. I implemented a workaround to handle this situation.
…tence There was a problem that we manipulate the words (merging tokens) and afterwards calling the CoreNLP to build a semantic graph out of it. This made it rather difficult to ask for the correct distances between tokens in the semantic tree. The new solution builds the graph right from the cleaned text (including MATH and LINK placeholders) and after merging words, they got the index of the first noun in the merged list of words or the index of the first word in the list if there is no noun present. Hence, 'jacobi polynomial' has the index for jacobi and not polynomial. When we calculate the semantic-graph distance now, it returns the distance to jacobi.
…graph generation from CoreNLP (obviously)
First, we allow to annotate just a single node in the moi dependency graph now rather than always annotating the entire document (thats useful for specific applications). Second the PosTagger only creates one instance of CoreNLP annotator per language and model path. So if you do not change the language or the model path in one run, there will be only one instance of CoreNLP running. Third, we load the dependency parser directly when instantiating NLP annotator. This allows us to build a dependency graph directly on annotation process and not afterwards. It also means we do not need to load a separate dependency parser from the filesystem (CoreNLP takes care of it now) which heavily reduces load time from 40s to 2s.
…writes the previous tags which results in differences between dependency graph and library of mathtags.
…f dependency graphs
Open
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Updates and patches the extractor to use MOI rather than single identifiers.