search

titipata / paper-reviewer-matcher

Linear programming solver for paper-reviewer matching and mind-matching
Apache License 2.0
82 stars 29 forks source link

Time estimate to run. #8

Open Franceshe opened 4 years ago

Franceshe commented 4 years ago

Hi, first thanks to open source the paper! I recently got interested in "matching algorithms" and run into this interesting paper. So I try to replicate the paper and run the sample as suggested. But it takes really long to run on my local mac(> 2.5 Hrs). You had referenced that "Here, we have around 500 users and recommended to trim around 50. This takes 1 hours to run." What's your benchmark for hardware for running this experiment?

Thanks!

titipata commented 4 years ago

Hi @Franceshe, I also use my Mac with touchbar first gen (15") to run the problem. I will look up for specifications tomorrow.

titipata commented 3 years ago

@Franceshe Oh boy, it takes a while for "tomorrow". So the hardware is Mac Touchbar 15" with 2.7 GHz Quad-Core Intel Core i7. Memory 16 GB.

I also added an example of mindmatch_cluster.py where we divide abstracts into clusters before performing mindmatching. That takes much shorter time to run.

Franceshe commented 3 years ago

@Franceshe Oh boy, it takes a while for "tomorrow". So the hardware is Mac Touchbar 15" with 2.7 GHz Quad-Core Intel Core i7. Memory 16 GB.

I also added an example of mindmatch_cluster.py where we divide abstracts into clusters before performing mindmatching. That takes much shorter time to run.

Yeah, it took a while but it is worthwhile lol. Thanks, I'll take try the example later and see how it goes :)

titipata commented 3 years ago

Awesome! Sorry that it took so long! I wrote my dissertation and that took forever, lol. I made a little trick where we do spectral clustering before running the mind-matching where you can run the following script.

python mindmatch_cluster.py data/mindmatch_example.csv --n_match=6 --n_trim=50 --n_clusters=4
Franceshe commented 3 years ago

Awesome! Sorry that it took so long! I wrote my dissertation and that took forever, lol. I made a little trick where we do spectral clustering before running the mind-matching where you can run the following script.

python mindmatch_cluster.py data/mindmatch_example.csv --n_match=6 --n_trim=50 --n_clusters=4

Looks like Covid is getting serious at Bangkok, Thailand. And looks like you had done so many interesting things in NLP and machine learning. Love your work, please take care hh.

titipata commented 3 years ago

Take care too! I will keep you update if I got the algorithm to scale in a better way.