# Deeper Text Understanding for IR with Contextual Neural Language Modeling

## TL;DR

This paper explores using BERT for deeper text understanding in information retrieval, demonstrating improved relevance modeling and effectiveness over traditional methods, especially with limited training data.

## Contribution

It introduces an enhanced BERT-based model for IR that leverages contextual understanding, outperforming traditional word embeddings and bag-of-words models.

## Key findings

- BERT-based representations outperform traditional word embeddings.
- Contextual models better leverage language structures in IR.
- Enhanced BERT benefits search tasks with limited training data.

## Abstract

Neural networks provide new possibilities to automatically learn complex language patterns and query-document relations. Neural IR models have achieved promising results in learning query-document relevance patterns, but few explorations have been done on understanding the text content of a query or a document. This paper studies leveraging a recently-proposed contextual neural language model, BERT, to provide deeper text understanding for IR. Experimental results demonstrate that the contextual text representations from BERT are more effective than traditional word embeddings. Compared to bag-of-words retrieval models, the contextual language model can better leverage language structures, bringing large improvements on queries written in natural languages. Combining the text understanding ability with search knowledge leads to an enhanced pre-trained BERT model that can benefit related search tasks where training data are limited.