This python code is used to evaluate coherence of topics from a topic model. This code first runs an API server using a pre-train word embedding model (app.py). Then use cm.py to evaluate the coherence metrics, average coherence or C@N (see Fang et al. 2016)

python app.py words.txt vectors.txt [port]

app.py takes 3 arguements:
words.txt is the vocabulary, corresponding to their word embedding vectors.txt.
[port] will be te port of this api server.

In a topic model, K topics are represented by the top 10 woprds, e.g. topics_exmaple.txt.
To evalute a single topic model:
python cm.py -f topic_exmaple.txt
To evaluate several topic models using the same configuration (e.g. K), put all topics file in a folder [topics_folders] and use the command:
python cm.py -f [topics_folders] -models

1. Using Word Embedding to Evaluate the Coherence of Topics from Twitter Data, Anjie Fang, Craig Macdonald, Iadh Ounis and Philip Habel. In Proc of SIGIR, 2016. pdf
2. Examining the Coherence of the Top Ranked Tweet Topics, Anjie Fang, Craig Macdonald, Iadh Ounis and Philip Habel. In Procs of SIGIR, 2016. pdf