Review_aspect_extraction
Review_aspect_extraction copied to clipboard
yafangy
CNN-based model to realize aspect extraction of restaurant reviews based on pre-trained word embeddings and part-of-speech tagging
Hi @yafangy Thanks for this nice attempt. There seems to be a problem with the output_text function in the Step3_evaluation_from_input.py file. The value of pred_y[count, wx] is 0 all through...
Hey is the aspect extraction for labelled or unlabelled data?
Hi @yafangy Great work. Thanks for sharing the code. I was preparing datasets for laptop. I need to know how did you decide **sent_len and sent_num** in case of restaurant?...
Hi, Does it work with languages other than English as it is and if not where can one make the necessary changes to make it work with other languages such...
Can the model be used for cross-domain prediction, i.e., domains other restaurants and laptops.
Step 1: Prep at line 169: create_train_data_restaurant(fn_train, args.out_dir+args.word_idx, args.out_dir, args.StanfordPOSTag_dir, args.domain,'Train', sent_len, sent_num) - I did try many days for searching this problem but it didn't help me to fix...
Step 3 - Evaluation from input: I got struggle with this line (line 273): dep_parser=StanfordDependencyParser(model_path="edu/stanford/nlp/models/lexparser/englishPCFG.ser.gz") Thanks for your reading!
script/Step1_prep.py searches for the .jar file in directory `stanford-postagger-full` but the dir is spelt as `stanford-posttagger-full`(with 2 t's in posttagger). renaming the dir appears to solve this.
Thanks for sharing the code. Is this model better than the original one (Double Embeddings )? Thanks John
