A leading provider of artificial intelligence, big data and data security products and services.

TRS independently developed the“TRS-GPT LLM Platform”and launched five industry-specific large language models for the Government Affairs,media, finance,public opinion, and public security.

The "White Paper" provides a comprehensive overview of various aspects of TRS's data elements, grounded in the practical experience of TRS's data element business.

TRS Information Retrieval Technology Adds National Invention Patents

Time: 2024-05-26
Recently, the research achievement entitled "Design of Inverted Index Structure for Segment Sentence Position and Its Limited Operation Full-text Retrieval Method" independently developed by Teres Information Technology Co., Ltd. has successfully obtained national invention patent authorization (Patent No.: ZL 2023 1 0847909.X), marking a further advancement in Teres' technical strength and innovative ability in the field of big data information retrieval. With this, the number of Teres' core patents in this field has reached 7, and the number of software copyrights has reached 21.
 
 
This invention innovatively proposes a method for designing an inverted index structure for segment sentence position and limiting its attribute operation for full-text retrieval, combining the construction of an inverted index list supporting segment sentence position attributes with attribute limited operations. By storing and involving limited operators as index attributes in the operation, it effectively solves the problem of users not being able to accurately query during the retrieval process due to structural defects in the existing inverted index technology, thereby improving the efficiency and accuracy of document retrieval.
 
For example, when a user wants to find articles published by the author Mo Yan, if they directly search with "author Mo Yan," the result set will contain many articles about "Mo Yan" written by other authors. However, by using segment sentence position retrieval and limiting the search of "author Mo Yan" within the same sentence range, the accuracy of search results can be significantly improved. This method can also be applied in specific application scenarios such as patent, thesis, report, news retrieval, etc., to better assist users in quickly locating the required information and improve retrieval efficiency.
 
TOS in the field of Chinese information retrieval has a history of 30 years of research and development. The heavyweight product TRS Haibei Search Database, rare in China, is a domestically developed search database from bottom-layer segmentation algorithms to full-text search engines and upper-layer management systems. Through multiple iterations, the current version of the Haibei Search Database integrates the TRS engine based on the positional inverted index structure. It has made significant breakthroughs in retrieval efficiency and retrieval effectiveness, suitable for text information retrieval scenarios with high precision requirements.
 
In the future, TRS will continue to focus on the research and application innovation of search core technology, providing users with a more powerful, higher-performance, and more stable domestically developed search database. This will help enhance the independent development and continuous strengthening of the search database industry, providing strong technical support for building a digital powerhouse.