Shuhua Liu Senior Research Fellow, Docent Arcada Universitty of Applied Sciences. KaTuMetro Kickoff Seminar, University of Helsinki

Transcription

1 Intelligent Methods and Models for Mining Community Knowledge: Enabling enriched Understanding of Urban Development in Helsinki Metropolitan Region with Social Intelligence Shuhua Liu Senior Research Fellow, Docent Arcada Universitty of Applied Sciences KaTuMetro Kickoff Seminar, University of Helsinki

2 Motivation and Objectives (1) Social media as potentially useful source of information and community knowledge relevant to urban development issues of Helsinki metropolitan region Modern day citizens generate large amount of information about where they are and what they are doing on social media, leaving marks and notes of their interaction with the urban environment. Conventional data More reliable, but expensive to collect, labour intensive, slow, do not scale easily Relatively sparse data, coarse location granularity Minimal context Social media data Cheaper, easier and faster to collect, massive amount, timely Geo-tagged data, fine-grained location data Richer demographics, more context Biases in data, Twitter bots, privacy implications

3 Motivation and Objectives (2) Regional development issues University of Helsinki, Dr. Maria Salonen, Prof. Tuuli Toivonen Cycling planning: understanding mobility patterns of people (summer vs winter), social media presence of cycling issues Green infrastructure and tourism resources Latest breakthrough developments in AI, deep learning methods and tools are empowering automated tools for social media content analysis. Objectives Exploring the potential and possibilities of using social media data to enrich our understanding of cities Developing methods, models and tools for extracting content and sentiment information from social media data Enabling innovative analytical approaches in urban planning

4 Research Questions, Tasks and Activities What types of information are potentially useful community knowledge relevant for the planning of cycling as well as green infrastructure and tourism resources in the region? information requirements and data collection How to automatically identify and extract the relevant parts and pieces of information from different social media content? methods, models and tools How to make use of the extracted information in urban planning and decision support? application cases How do the models perform? What are the pros and cons of the methods, models and tools? testing and validation Close collaboration with University of Helsinki

5 Social Media Content Analysis using AI Methods LDA Topic modelling and visualization Discover hidden thematic structure in large text collection, extract topics Sentiment/attitude analysis Lexicon-based and learning based sentiment analysis Entity extraction, Event extraction Deep Learning models and tools Recurrent NN, TensorFlow Transfer learning built upon successful deep neural network models Different models for different types of source data and different information to be extracted Handling Finnish and Swedish text Joint analysis of text and image data

6 Outputs, Dissemination and Potential Impacts State-of-the-art research results, international conferences, open source software prototypes Project meetings, workshops/seminars, social media, web sites, to engage our stakeholders and wider audience Bring advanced data technology developments and innovations closer to urban development and to public decision making. Hope to open up new possibilities for our stakeholders in enriching and enhancing decision process in urban planning, and developing better understanding of regional resources and development issues; also useful for companies in the region Contribute to the intelligent infrastructures needed to build smart city services

7 Thanks!

8