cttsai/illinois-cross-lingual-wikifier

illinois-cross-lingual-wikifier's Issues

I can not get some of the data, can you help me?

I have already obtained xlwikifier-mapdb, xlwikifier-data; but tac2016_es_eval, tac2016_en_eval, tac2016_en_eval, tac2016_eval_golds how to get the configuration file?

bug in surfaceclustering

The loop where this occurs:
https://github.com/cttsai/illinois-cross-lingual-wikifier/blob/master/src/main/java/edu/illinois/cs/cogcomp/xlwikifier/postprocessing/SurfaceClustering.java#L170
appears to be incorrect. The size of the c1 and c2 surface form sets is determined by how many different surface forms there are; so the clustering threshold appears to allow unification of clusters ONLY when there are num_threshold distinct surface forms, whereas it seems as though it should be due to cluster size. Is that correct?

mention label scores can have null labels

the map from label to score used internally to track targets, and which is now mapped into Constituents' labelToScore map, can contain null labels. It would make life simpler for users if only non-null labels were permitted.

demo not working; link to dataset is dead

Hi,

First of all I'd like to say the project looks interesting, thanks for sharing it!

However the demo https://cogcomp.seas.upenn.edu/page/demo_view/xl_wikifier seemed to have JS errors; and the dataset download http://cogcomp.cs.illinois.edu/Data/ccgPapersData/ctsai12/xlwikifier-mapdb.tar.gz is now returning 404

Are you interested in fixing these? Also btw, I am reading the paper Cross-lingual Wikification Using Multilingual Embeddings and it was 4 years ago, any further researches you have done on this topic?

Thanks very much!

Problem in Wikification

Hello,

I am using the Wikifier tool, but I am getting an error as image attached in the email below. The file D4 is in the folder but it's not able to read it, I am facing a similar issue with 200 files out of 600 files in the folder.

I am using WIkiQA dataset, which talks about the different question in each document so Can I pass all the data together to the Wikifier or do I need to give one by one (for a particular topic like, president of the USA, global warming etc.)?

Please share your opinion.

should fail if no db found

in ConfigParameters, should fail catastrophically with an informative error message if the database is not found.