dice-group / Palmetto

Palmetto is a quality measuring tool for topics
GNU Affero General Public License v3.0
209 stars 36 forks source link

Parallel computation of multiple coherences #77

Open MichaelRoeder opened 1 year ago

MichaelRoeder commented 1 year ago

User story

As a researcher, I am interested in calculating several coherence values for a given topic. At the moment, I have to rerun Palmetto several times. Each time, I provide the set of all topics and one of the coherence measures for which I would like to get the values. I would appreciate it if I could simply provide a list of coherence measures and Palmetto would give me all the results at once.

Another advantage would be the time that could be saved. Since the index is the performance bottleneck, it might be faster to simply calculate all coherence values for a single topic while the necessary counts are in the main memory instead of reading them again and again from the index.

Development steps