The Smart City Project

In this project, we plan to build a people-aware smart city framework, which focuses on finding people’s needs and satisfy them. This is the first attempt to build such a smart city framework that centers on people, the residents of a city, we proposed a people-aware smart city framework that integrates data extracted continuously from the people, discovers their needs from integrated multi-source data, and finally determines the best resource allocation plans to satisfy these needs. 

People’s needs from the areas of education, health, travel, safety, finance and entertainment, which all have measurable objectives, will be studied in this project. To achieve the goals in the framework, several state-of-the-art techniques will be developed including data integration solutions to handle different data sources with different formats, transfer learning-based mechanisms to reveal knowledge, and machine-human collaborative approaches to make wise decision. In addition to making breakthrough in technical development of people-aware smart city framework, we will closely work together with our sponsor and partner, China Digital City Forum Limited, to implement our framework into their smart city solutions and demonstrate the effectiveness.

We have applied data extraction techniques for different data sources, such as Web and social media. For data extraction, we have built tools to extract various kinds of data from a variety of websites.

For similar data from different sources, such as financial news, job recruitment, and real estate information, we have extracted the commonalities between the different sources, integrate these data, and designed a unified data structure to store the corresponding information from various sources.

We have conducted three steps based on the granularity of integration: schema mapping, record linkage, and data fusion. Schema mapping in a data integration step refers to (i) creating a mediated (global) schema, and (ii) identifying the mappings between the mediated (global) schema and the local schemas of the data sources. For the record linkage, both the machine-based automatic technique and crowd-based technique have been developed based on the features of string similarity, attribute overlapping and so forth. The basic idea is to map the both structured and unstructured data to a knowledge base and then perform the linkage on the KB, the mapping process could be assisted by the crowd. Data fusion refers to resolving conflicts from different sources and finding the truth that reflects the real world. Therefore, how to find all the true values while filtering out the noise values isquite challenging. To make this process more scalable, we consider the dependency rules and logical inference patterns at the same time.

To facilitate data integration, we have implemented a system to output all the extracted data in similar JSON formats according to a predefined standard. We also provided an API for programmers to retrieve the data. We have also already implemented some web applications like daily transaction history search engine, which provides services to the public by integrating the data we have extracted.