Word Segmentation in Sanskrit Using Energy Based Models
Lists
Only you see this
Abstract
This is the repository for word segmentation in sanskrit using energy based models. # Word Segmentation in Sanskrit Using Energy Based Models<br> <br> <br> ## Getting Started<br> <br> Please download the 2 compressed files 'dir.zip' and 'wordsegmentation.rar' to your working directory and extract them into folders named 'dir' and 'wordsegmentation' respectively.<br> <br> Your working directory should be as follows<br> * Working Directory<br> * wordsegmentation<br> * skt_dcs_DS.bz2_4K_bigram_mir_10K<br> * skt_dcs_DS.bz2_4K_bigram_mir_heldout<br> * dir<br> <br> ## Prerequisites<br> * Python3<br> * scipy<br> * numpy<br> * csv<br> * pickle<br> * multiprocessing<br> * bz2<br> ## Instructions for Training<br> Change your current directory to 'dir'<br> <br> Run the file Train_clique.py by using the following command<br> <br> * python Train_clique.py<br> <br> To train on different input features like BM2,BM3,BR2,BR3,PM2,PM3,PR,PR3 please modify the bz2_input_folder value in the main function before beginning the training.<br> <br> Feature | bz2_input_folder<br> ------------- | -------------<br> BM2 | wordsegmentation/skt_dcs_DS.bz2_4K_bigram_mir_10K/<br> BM3 | wordsegmentation/skt_dcs_DS.bz2_1L_bigram_mir_10K<br> BR2 | wordsegmentation/skt_dcs_DS.bz2_4K_bigram_rfe_10K/<br> BR3 | wordsegmentation/skt_dcs_DS.bz2_1L_bigram_rfe_10K/<br> PM2 | wordsegmentation/skt_dcs_DS.bz2_4K_pmi_mir_10K/<br> PM3 | wordsegmentation/skt_dcs_DS.bz2_1L_pmi_mir_10K2/<br> PR2 | wordsegmentation/skt_dcs_DS.bz2_4K_pmi_rfe_10K/<br> PR3 | wordsegmentation/skt_dcs_DS.bz2_1L_pmi_rfe_10K/<br> <br> ## Instructions for Testing<br> <br> After training, please modify the 'modelList' dictionary in 'test_clique.py' with the name of the neural network that has been saved during training. While testing for a feature, please provide the name of the neural net which was trained for the same feature.<br> <br> We only provide the trained model for the feature BM2 which was our best performing feature. If the name of the neural net is not changed, then the testing will be performed on the pre-trained model for BM2 provided in outputs/train_t7978754709018<br> <br> To test with a particular feature vector use the tag of the feature while execution<br> <br> * python test_clique.py -t <tag><br> <br> For example: <br> * python test_clique.py -t BM2<br> <br> After finishing the testing please run the following command to see the precision and recall values for both the word and word++ prediction tasks<br> <br> * python evaluate.py <tag><br> <br> For example: <br> * python evaluate.py BM2
BibTeX
@inproceedings{krishna2018word,
author = {Amrith Krishna and Bishal Santra and Sasi Prasanth Bandaru and Gaurav Sahu and V. D. Sharma and Pavankumar Satuluri and Pawan Goyal},
title = {{Word Segmentation in Sanskrit Using Energy Based Models}},
booktitle = {Zenodo},
year = {2018},
doi = {10.5281/zenodo.1035413},
}Loading related content...