Update address locator service with AI Semantic analysis

Talk about “AI” what will you think of? In addition to the voice robot which saying that “Do you need help? ”, I have no idea what are you talking about.” AI artificial intelligence can be applied beyond your imagination.

SuperGeo collaborates with and GOLiFE to develop the software of Address Locator service to translate the address to space coordinates. However, because the vocabulary of the address is complicated and the structure is not completely unified and often maintained by different offices, which is a lack of consistent standards, it causes difficulties in analysis and thus affects the positioning accuracy. Therefore, SuperGeo tries to optimize the address software by introducing AI to get more accurate positioning information.

The project, Update address locator service with AI Semantic analysis, is supported by the AI value-added transformation application of the information service industry promotion products promoted by the Industrial Development Bureau, Ministry of Economic Affairs.

Solution and benefit

Team SuperGeo tried to introduce artificial intelligence technology to solve the technical bottleneck of Address Locator products. By using two methods of natural language processing technology: “Address Normalization” and “Chinese word-breaking tool” (AI machine learning model), to optimize the address location capability.

To deal with the various forms of the actual address, one of the ways to evaluate the accuracy of address location is to use the concept of Address Normalization, which restricts the form of the address by a specific specification, so as to facilitate the subsequent use of the data. However, it is still difficult to generalize the ambiguity of all addresses by artificially formulated logic in the programming language. Therefore, the AI technology is introduced, and the machine learning algorithms provided by the Chinese word segmentation tool “HanLP” ,such as hidden Markov model, conditional random field model and perceptron Model, using the address element corpus as the model training target, then establishing an address segmentation model to assist the address to be correctly split and successfully positioned.

After the product value is added, it can support more information patterns, improve the ability of address comparison coordinates, reduce the time cost of manual processing, lower the threshold of address location technology, and help users to attain spatial information from a large amount of business, thus inducing deep into various spaces, including the use of government open data to provide public information, life safety and quality, elderly nursing, transportation and communication, recreation and other problem-oriented solutions, opening the possibility of entering geospatial analysis.

Source: https://www.supergeotek.com/tw/index.php/aipoisition/


More Posts

2022 High Level Political Forum (HLPF)

Hosted by the United Nations brings the leaders and civil society organizations worldwide together every July to examine the status of the 2030 Sustainable Development Goals (SDGs) and propose solutions.