Geographic Semantics-Based Quality Evaluation Method for Place Name and Address Retrieval

Abstract

The invention discloses a geographic name address retrieval quality evaluation method based on geographic semantics, is applicable to the problem of evaluation of geographic name address retrieval quality, and provides a quantitative evaluation scheme of the geographic name address retrieval quality. The method includes the steps: inputting a geographic name address, and processing the inputted geographic name address by a geographic name address retrieval engine to generate a retrieval result; analyzing the geographic name address by an evaluation engine based on the geographic semantics to generate a geographic name address word vector and a standardization address unit, and transmitting the geographic name address word vector and the standardization address unit into a text similarity evaluator and an address granularity evaluator; calculating the similarity of the inputted address and the retrieval result by the text similarity evaluator by the aid of a word vector computing technique according to a geographic name word spectrum, and returning the similarity to a comprehensive evaluator; performing surplus deletion and few supplementing on the standardization address unit by anaddress granularity evaluator based on a geographical name address standard model, calculating a minimum granularity index of the inputted address, and returning the minimum granularity index to thecomprehensive evaluator; calculating an evaluation value of the retrieval result by the comprehensive evaluator according to the similarity of the word vector and the minimum granularity index of theaddress.

Description

technical field [0001] The invention belongs to the field of geographic information systems, and proposes a geographic semantics-based retrieval quality evaluation method for place names and addresses. Background technique [0002] With the advent of the era of big data, spatial location information, as an important dimension of urban information, has received more and more attention. In the business data owned by functional units such as urban industry and commerce, taxation, planning, public security, and banks, spatial location information such as areas, streets, and house numbers are usually described in text form. Therefore, more and more place-name-address retrieval engines that map place-name addresses into latitude and longitude are put into use. In the face of many search engines for place names and addresses, how to evaluate the quality of search results for place names and addresses, and what perspective and method should be used to make an accurate and fast eval...

AI Technical Summary

Problems solved by technology

This kind of method technology is relatively simple and convenient, but it is impossible to make an accurate description of the geographical location of the entity, and then it is impossible to make a quantitative evaluation of the retrieval results of place names and addresses

Another angle is from the perspective of the address encoding algorithm, the retrieval quality is evaluated by matching the location accuracy of the address, while ignoring the semantic similarity of the text, so this evaluation method is not perfect

Images