dhruvjain/app_ir_project

JavaScript

ACL TimeLine Retrieval

Timeline retrieval of academic papers from ACL Anthology collection

References

Slides for Presentation

[Slides] (https://docs.google.com/presentation/d/1oJz0S1t27yeFW5U0F750fuiW3Oh97I41tP352O_HgQ0/edit?usp=sharing)

Instructions to use

TP - ICP

run python doc_tp.py to get all the topics and get topic proportions. This will create doc_tp_scores.npy object.
run python doc_tp_icp.py to get TP-ICP scores for all the documents. This will create ICP.npy and TP_ICP_DOC.npy object.

Document Similarity

run python doc_similarity.py to get similarity between documents. This will create DOC_SIMILARITY.npy object.
run python doc_similarity_normalised.py to get similarity between documents. This will create DOC_SIMILARITY_NORMALIZED.npy object.

Random Walk with restarts

run python random_walk_parallel.py to get similarity between documents. This will create 9916 numpy objects based on random walk being started from each node.

Running the Web Application

run python manage.py runserver 0:8000 to start the server. Open your browser and goto 127.0.0.1:8000

Requirements

NumPy
Django

Sample Screenshots

Ranked retrieval

Year-wise Retrieval

Clustered Retrieval

Diverse Retrieval