Somali Information Retrieval Corpus: Bridging the Gap between Query Translation and Dedicated Language Resources

This is the official GitHub for the paper: Somali Information Retrieval Corpus: Bridging the Gap between Query Translation and Dedicated Language Resources

In this corpus we have listed the important components as:

1. Somali IR Evaluation Corpus: holds the documents, with 2335 Somali text documents. Rename this file as Somali_IR_Evaluation_Corpus
2. Judgements: the number of documents judged against the total documents concerning the queries.
3. queries.txt: holds the queries used 16 in total.
4. Somali_Stop_Words.txt: the number of Somali stop words identified are in this file
5. Som_PRF_IR.py: this file contains the Psuedo relevance feedback technique implementation
6. Som_TF-IDF.py: the TF-IDF python file is in this file
7. PRF-Evaluation: Results after Som_PRF.py is run.
8. TF-IDF-Evaluation: holds results after successfully running Som_TF-IDF.py
9. Som_index_creator.py: if you want to re-run the whole process this file creates the index, you need to run it first.
10. evaluation.py: is the file that holds scripts used to evaluate the performance.

Usage

Please run the Evalaution.py file and exchange the two file names that hold the values of the two models. If you want to use the TF-IDF the file name must be TF-IDF-Evaluation and if you want to use the PRF the file name must be PRF-Evaluation, these two files are inside the Evaluation.py.

After downloading please change the folder name to Somali_IR_Evaluation_Corpus