search

swisscom / ai-research-keyphrase-extraction

EmbedRank: Unsupervised Keyphrase Extraction using Sentence Embeddings (official implementation)
Apache License 2.0
432 stars 88 forks source link

Train new model for arabic and persian language #27

Closed mahaghdam closed 4 years ago

mahaghdam commented 4 years ago

Hello I run the module with pretrained model, it works fine. I want to train new models for arabic and persian. I trained sent2vec model but I didn't work. Can you please guide me what else should I do? How can I use this module for other languages?

mahaghdam commented 4 years ago

Hello again I used stanfordpostagger to tag the input text for my aimed languages. Convert different tags to considered tags. But I still have problems with extracting keyphrases. There is nothing as outout. Can you please guide me what I'm missing?

kamilbs commented 4 years ago

Hi ! Did you create a regular expression (based on pos tags) to extract candidate phrases ? I'll suggest you to check that you have rule(s) that extract candidate phrases in your language. extractor.py