PROJECT TITLE :

Traffic Information Mining From Social Media Based on the MC-LSTM-Conv Model

ABSTRACT:

The use of social media such as Sina Weibo has the advantage of reflecting traffic information. This information can include the reasons for traffic jams, illegal behaviors, and emergency recourses on roads. Nevertheless, there is still the difficult problem of how to mine traffic information sufficiently. In this paper, we propose a method for the management of traffic jams that is based on Deep Learning and makes use of data from social media. The method that is being proposed is based on two primary ideas. First, a multichannel network known as MC-LSTM-Conv is proposed. This network will have a layer for Long Short-Term Memory (LSTM-layer) as well as a layer for Convolution (Conv-layer). Two information channels have been incorporated into this model in order to derive abstract characteristics from input text. Each channel has two Conv-layers, and one of the four Conv-layers also has an LSTM-layer added to it. From large amounts of Sina Weibo data, check-in microblogs that reflect traffic jams are extracted with the help of a model called MC-LSTM-Conv. In the second step of the process, a set of matching rules is developed by basing them on the keywords that are associated with images of traffic congestion. These rules further categorize the microblogs that were extracted in the first step into a total of four classes, with each class representing a different kind of road condition (i.e., traffic accidents or large-scale activities, road construction, traffic lights, and the low efficiency of government agencies). The performance of the proposed multichannel network in extracting microblogs about traffic jams is demonstrated experimentally using data from Sina Weibo. The method of keyword fuzzy matching is an effective way to retrieve detailed information concerning traffic jams.