Ankita Sharma

Title: Ankita Sharma and Her Work in Information Retrieval.

Introduction

Ankita Sharma is an inventor based in Sunnyvale, California. Her patent work addresses information retrieval, with a focus on organizing document representations to improve how systems find relevant information.

Latest Patents

Sharma is associated with 1 patent, titled “Information Retrieval System Using a Hierarchical Corpus Encoder.” The invention adapts a dense encoder into a hierarchical corpus encoder. It uses negative samples from sibling nodes in a tree of document embeddings, while jointly training the encoder and the hierarchy with a loss function that accounts for document relationships.

The approach can support supervised training, when query-document relevance judgments are available, as well as zero-shot use, when a query dataset is absent. The patent describes performance improvements over several dense-encoder and generative-retrieval baselines in both supervised and unsupervised settings, highlighting the potential value of jointly learning a document hierarchy.

Career Highlights

Sharma is associated with Microsoft Technology Licensing, LLC. Her patent reflects work on machine-learning methods for information retrieval and the organization of large document collections.

Collaborations

Sharma’s listed coworker is Tongfei Chen. Their association connects her patent work with other contributors in the information-retrieval field.

Conclusion

Ankita Sharma’s patent presents a hierarchical approach to document encoding that aims to improve retrieval across different training scenarios. It represents her contribution to advancing machine-learning techniques for finding relevant information.