Embedding based approach over Traditional LDA for AdHoc Retrieval

[bhavan-bhatt]

Title

Embedding based approach over Traditional LDA for AdHoc Retrieval

Team Name

Retrieval Kings

Email

202311021@daiict.ac.in

Team Member 1 Name

Bhavan Bhatt

Team Member 1 Id

202311021

Team Member 2 Name

Ayush Hirdani

Team Member 2 Id

202101139

Team Member 3 Name

Harshneel Soni

Team Member 3 Id

202311024

Team Member 4 Name

Samarth Motka

Team Member 4 Id

202311023

Category

Reproducibility

Problem Statement

Existing systems utilising LDA for ad-hoc retrieval only look at words on a global level, there have been improvements on it using Word2Vec we intend on reproducing the results and then improving it by using other techniques.

Evaluation Strategy

MAP, GMAP and Recall

Dataset

Robust - https://trec.nist.gov/data/robust/04.guidelines.html

Resources

Title - LDA-Based Document Models for Ad-hoc Retrieval Link - https://dl.acm.org/doi/10.1145/1148170.1148204

Title - A Word Embedding based Generalized Language Model for Information Retrieval Link - https://dl.acm.org/doi/10.1145/2766462.2767780

[parth126]

• Problem needs to be a little deeper (try modifying the approach in some manner) or broader (experiment with more language models)