Patent retrieval method and device based on multiple rounds of conversations

Abstract

The invention discloses a patent retrieval method and device based on multiple rounds of conversations, and the method comprises the steps: obtaining an input statement of an input side, and carryingout the word segmentation of the input statement to generate a plurality of segmented words; performing named entity identification processing on the plurality of segmented words to obtain a pluralityof entities corresponding to the input statement; performing intention detection on the plurality of entities to obtain an intention result corresponding to the input statement; acquiring historicaldialogue information of an input party, and updating current state information according to the historical dialogue information and the intention result; obtaining an execution action corresponding tothe updated current state information, and generating a query request corresponding to the input statement according to the execution action; and inputting the query request into the patent knowledgebase to obtain a query result, adding the query result into a preset template, and returning the query result to the input party. According to the method, the current retrieval result is presented ina multiple rounds of conversations mode, the user is assisted in carrying out patent retrieval requirements, the requirement description of the user can be standardized, and the accuracy of patent retrieval is remarkably improved.

Description

technical field [0001] The invention relates to the technical field of information interaction, in particular to a patent retrieval method and device based on multi-round dialogs. Background technique [0002] With the vigorous development of science and technology in our country, "mass entrepreneurship and innovation" has become the source of power. In 2018, the number of invention patent applications in China was 1.542 million. A total of 432,000 invention patents were authorized, including 346,000 domestic invention patents. As one of the main presentation forms of scientific research achievements, patents are an important reference and reliance for scientific and technological development and transformation of achievements. However, with the rapid expansion of the patent resource pool, how to search for patents that match one's own needs in the huge resource pool has become an important issue in the transformation of scientific and technological achievements. At presen...

AI Technical Summary

Problems solved by technology

[0007] Solution 1) is basic and low in complexity. It is highly efficient for professionals who are proficient in patent retrieval, but it has high requirements for users. For most ordinary users who have no experience in patent retrieval, it is difficult to convert their own needs into accurate keyword search, which makes it difficult to search for patents that meet the needs

Solution 2) The feature extraction of patents can be done offline, so the retrieval complexity is low, and the requirements for users are low. You only need to input the keywords required by the patent, and the operation is simple, but a lot of detailed information will be lost in the process of extracting patent features. Therefore, the accuracy of query results is difficult to guarantee

Solution 3) Advanced technologies in the field of natural language processing such as semantic analysis and content retrieval are used to better understand the patent retrieval requirements input by users and return matching patents. The machine learning algorithm model adopted can be trained offline, so patent retrieval is complicated Low accuracy and low requirements for users; however, intelligent question and answer lacks context information, and users must input all requirements at once. When users cannot clearly describe the requirements, the accuracy rate is difficult to guarantee

[0008] The above-mentioned three current mainstream patent retrieval schemes still have a common problem, that is, they cannot provide feedback on the user's demand description. For example, when the user's demand description deviates from the mainstream and accurate description, the user cannot be prompted and confirmed. , affecting the user's final search results

Images