Semantic Language Prediction Models
- Technology Benefits
- - Achieve significantly better translation quality measured by the BLEU score- Improves readability of the translation
- Technology Application
- - Machine translation- Speech recognition programs/applications - Information retrieval
- Detailed Technology Description
- None
- Supplementary Information
- Inventor: Wang, Shaojun | Tan, Ming
Priority Number: US20130325436A1
IPC Current: G06F001727 | G10L001518
US Class: 704009 | 704E15018
Assignee Applicant: Wright State University,Dayton
Title: Large Scale Distributed Syntactic, Semantic and Lexical Language Models
Usefulness: Large Scale Distributed Syntactic, Semantic and Lexical Language Models
Novelty: Composite language models for machine translation, speech recognition and information retrieval, has composite word predictor to predict next word, and approximate expectation-maximization and follow-up expectation-maximization algorithm
- Industry
- ICT/Telecom
- Sub Category
- Software/Application
- *Abstract
-
The inventors have created an algorithm which builds powerful language models to capture and predict natural language. The invention has a variety of real world applications involving machine translation, speech recognition software and information/ database retrieval systems. The improvements provided by the invention are done without sacrificing strong assumptions or intractable model assumptions. A working language model is available via the Ohio Supercomputer Center.
- *Principal Investigator
-
Name: Ming Tan
Department:
Name: Shaojun Wang
Department:
- Country/Region
- USA
For more information, please click Here