Servicenow, Inc.

Microsoft Technology Licensing, LLC

Srivatsava Daruru

Quaizar Vohra

Mitul Tiwari

Madhusudan Mathihalli

Kaushik Rangadurai

Ravi Narasimhan Raj

Deepak Bobbarjung

Abhisaar Yadav

Rajeev Kumar

Silviu-Petru Cucerzan

Anil Kumar Madamala

Jignesh Parmar

Olivier Nguyen

Ziaul Hasan Hashmi

Deepak Zambre

Orlando Marquez

Patrice Bechard

Rohit Paravastu

Shounak Purkayastha

Soham Parikh

A method is provided for efficiently providing sentiments or other manual labels for textual training data. The method includes using an embedding model to project acquired user text to an embedding vector in an embedding space. Distances (e.g., cosine similarities) between this embedding vector and the embedding vectors determined for a plurality of already-label user text training examples are then determined. The already-labeled user text that has the shortest distance is determined and the label thereof is prospectively applied to the acquired user text and presented to a user for approval. The user can approve the prospectively applied label, in which case the newly acquired text is added to the training data with the prospectively applied label associated therewith for later use in training a language model. Alternatively, the user can decline the prospectively applied label and apply an alternative label to the newly acquired text.

Method for filtering and semi-automatically labeling training data

Example data processing systems and methods are described. In one implementation, a system accesses a corpus of data and analyzes the data contained in the corpus of data to identify multiple documents. The system generates vector indexes for the multiple documents such that the vector indexes allow a computing system to quickly access the plurality of documents and identify an answer to a question associated with the corpus of data.

Generation and utilization of vector indexes for data processing systems and methods

Data processing systems and methods

Representative embodiments disclose mechanisms to validate statements made as part of a document creation or editing process. As the user edits or creates a document using a document creation/editing application, statements made in the document can be submitted to a validation service in a proactive or reactive manner. The statement validation service receives the statement, resolves any coreferences using a coreference resolution process. Once coreferences in the statement have been resolved and replaced, entities are resolved using an entity resolution process. Predicates are then resolved using a predicate resolution process. Entity-predicate pairs are then used to traverse a knowledge graph to extract information relevant to the statement. Suggested corrections are created from the extracted information and presented to the user via the document creation/editing application.

Fact validation in document editors

Growing community of inventors

Mountain View, CA, United States of America

Srivatsava Daruru