Tree-structure-based question answering system and method

Abstract

The invention discloses a tree-structure-based question answering system and method. A tree structure is built according to correlation of all question service attributes and different levels of all questions under services that the questions belong to; each node of the tree structure stores the questions and corresponding answers; each node of the tree structure is provided with a node identifier. The node questions and node identifiers in the tree structure are displayed to a client; after the node identifiers transmitted by the client over internet are received, the questions stored in the nodes corresponding to the node identifiers and the corresponding answers are supplied to the client. The tree-structure-based question answering system and method provides questions of the client with more accurate answers, so that user experience of the client is improved.

Description

technical field [0001] The invention relates to the field of electronic commerce, in particular to a method and system for answering questions based on a tree structure. Background technique [0002] E-commerce mainly uses the Internet to engage in business or activities. In the process of e-commerce, e-commerce companies need to set up online customer service on the network side to provide online answers to various business-related questions for clients through the Internet. Consultation on related attributes of business products, consultation on related activities of e-commerce enterprises and after-sales service consultation, etc. With the increasing number of inquiries from clients, the limited online customer service can no longer meet the needs. In order to meet the increasing number of inquiries from clients, a response system is set up on the network side to automatically reply to various business-related questions raised by clients. [0003] When the response syste...

AI Technical Summary

Problems solved by technology

However, on the one hand, when the question is expressed in Chinese, its semantic recognition is more difficult. The meaning expressed by the same phrase in different pragmatics and contexts is completely different. To recognize the meaning expressed in Chinese, it is necessary to convert a sentence to Words are divided into several words. Because Chinese has a very strong ambiguity, none of the existing analysis techniques can effectively divide a sentence.

On the other hand, the above two schemes use the inverted index and one-by-one matching method to match questions and corresponding answer information when realizing the answer of the answering system. This matching method does not take into account the correlation between questions. Instead, the query information is directly matched with the question, so the matched question may not be what the client needs, and the corresponding response information is not the response information that the client needs, resulting in a lower matching degree

Images