Tongyu Zhu

The scaling of human mobility by taxis is exponential

As a significant factor in urban planning, traffic forecasting and prediction of epidemics, modeling patterns of human mobility draws intensive attention from researchers for decades. Power-law distribution and its variations are observed from quite a few real-world human mobility datasets such as the movements of banking notes, trackings of cell phone users’ locations and trajectories of vehicles. In this paper, we build models for 20 million trajectories with fine granularity collected from more than 10 thousand taxis in Beijing. In contrast to most models observed in human mobility data, the taxis’ traveling displacements in urban areas tend to follow an exponential distribution instead of a power-law. Similarly, the elapsed time can also be well approximated by an exponential distribution. Worth mentioning, analysis of the interevent time indicates the bursty nature of human mobility, similar to many other human activities.

A Heuristic Map-Matching Algorithm by Using Vector-Based Recognition

The traditional map-matching algorithms mainly use two methods: the incremental method and the global method. Both of them have advantages and disadvantages of themselves: while the global map-matching algorithm produces better matching results, the incremental algorithm produces results of lower quality faster. All things considering the two traditional algorithms, this paper proposes a heuristic map-matching algorithm by using vector-based recognition. Firstly, the algorithm uses the heuristic search method which is similar to A* algorithm, and it makes use of geometric operation to form the restriction, and make the comparison between the vector formed with the vehicular GPS points and the special road network to heuristicly search and select the vehicular possible traveling routes. Secondly, it globally compares the vehicular every possible route by calculating the map-matching weight, and then chooses the optimal one. The result of testing demonstrates the efficiency of the algorithm both at accuracy and computational speed when handling the large-scale data of GPS tracking data even under the complex road network conditions.

Outlier mining based Automatic Incident Detection on urban arterial road

Nowadays, Floating Car Data (FCD), which is becoming an important way to acquire traffic information, has been widely taken to estimate speed or travel time on road. In this paper, we introduce the concept of outlier mining into Automatic Incident Detection (AID) based on FCD and propose a novel AID approach on urban arterial road. According to the characteristics of incident, feature vector is selected from both spatial analysis and temporal analysis. Then a multilevel detection method that consists of filtering, outlier detection, and delay monitoring is proposed. The evaluation on real incident data and FCD gives the result that DR = 81.5% while FAR = 1.83%, which proves that the approach can achieve considerable effectiveness.

The bus arrival time service based on dynamic traffic information

The bus arri val time (BAT) service is a key service to improve public transport attractiveness by providing use rs with real-time bus arrival information which can help them to arrange their bus travel schedule intelligently. Thus the technique of real-time bus arrival prediction has become a research hotspot in the community of Intelligent Transport Systems (ITS) nowadays. In this paper, a novel model on bus arrival time prediction is proposed. The model proposes a complete set of programs to solve BAT prediction for large-scale real-time traffic information calculating. It adopts an effective algorithm judging buss driving direction real-timely. BAT is calculated based on dynamic traffic information and visual prediction is a way to complement when GPS information is not arrived as expected. Experimental results indicate that the model has considerable efficiency in accuracy (over 85.1%) and computational speed (max 5000 GPS records per second).

A map matching algorithm for intersections based on Floating Car Data

The traditional map matching algorithms consider little about the complicated structure of the road network, and regard all roads as the same. However, in the transportation information system using the floating car data (FCD), the GPS sampling rate is low, and it is probable to figure out the incorrect result when the vehicle is in the intersection area. To solve this problem, this paper proposes a bidirectional heuristic map matching algorithm for intersections based on a data structure for intersections. This algorithm apply the data structure of intersections to separates the intersection part from common map matching, decreases the FCD map matching mistakes that caused by the complicated road network and the GPS errors, and increases the accuracy of map matching.

Missing Data Compensation Model in Real-Time Traffic Information Service System

Nowadays the floating car data (FCD) is playing a more and more important role in the route guidance, because it can collect more accurate travel time information in traffic service systems. But the problem is that the congested traffic performs to be dynamic and quite complex, and therepsilas great data amount fluctuation at different time of day. Therefore, the stability of the data canpsilat be guaranteed, and the OD travel time calculated in dynamic route guidance will be seriously affected. In this paper, a multipattern compensating model for blank-data links which is based on the optimized road network is proposed. According to the classified historical data, the missing real-time traffic data is filled up. Finally the experiment which involved the data of 15,000 taxies for 6 months was carried out in several ways. The result suggests that this method raised the road coverage while guaranteed the accuracy and it can be applied in real-time systems to manage large amount of data.

Combining long-term and short-term user interest for personalized hashtag recommendation

Hashtags, terms prefixed by a hash-symbol #, are widely used and inserted anywhere within short messages (tweets) on micro-blogging systems as they present rich sentiment information on topics that people are interested in. In this paper, we focus on the problem of hashtag recommendation considering their personalized and temporal aspects. As far as we know, this is the first work addressing this issue specially to recommend personalized hashtags combining longterm and short-term user interest.We introduce three features to capture personal and temporal user interest: 1) hashtag textual information; 2) user behavior; and 3) time. We offer two recommendation models for comparison: a linearcombined model, and an enhanced session-based temporal graph (STG) model, Topic-STG, considering the features to learn user preferences and subsequently recommend personalized hashtags. Experiments on two real tweet datasets illustrate the effectiveness of the proposed models and algorithms.

Continuous range monitoring of moving objects in road networks

A set of distributed continual range query requests, each defining a geographical region of interest, needs to be periodically reevaluated to provide up to date answers. Processing these continual queries efficiently and incrementally becomes important for location based services and applications. In this paper, we propose an efficient incremental method for continuous range query characterized by mobility of objects which follow paths in an underlying spatial road network. We notice that both queries and moving objects can be attached to the road link for updating of query answers. By using Grid R∗-tree which indexing spatial road network data, query requests and moving objects register themselves to corresponding road links according to query areas and moving object locations on the network. Also we analyzed the impacts to query answers caused by geometry relation of road link and query area and state of moving objects. Then we proposed an efficient update algorithm and corresponding memory data structures based on spatial and state classification which formed a safe region to avoid unnecessary updates. Finally a comprehensive experimental evaluation using real data has been conducted to demonstrate the efficiency and effectiveness of our algorithm.

A history data based traffic incident impact analyzing and predicting method

Traffic incidents are a main factor that reduces capacity and service quality of roads. Due to the absence of efficient incident impact analyzing and predicting method, the traffic congestion and secondary accident brought by traffic incidents can hardly be avoided. In this paper we propose a history data based traffic incident impact analyzing and predicting method. By extracting regularity and volatility information from history data, we find a solution to analyze both parts that comprise the traffic flow status: the road condition without the incident, and the impact of the incident. Thereby we can predict the traffic condition under incidents by estimating the two components separately and adding them together. Experimental results show that our solution could simulate and predict the impact tendency of traffic incidents with high accuracy.

An effective method to estimate urban link travel time in real-time traffic information system

Nowadays the Floating Car technology is playing a more and more important role in the dynamic route guidance, congestion management, traffic incidents detection, because it can collect more accurate travel time information in real-time traffic service systems. However the key problem of using it is that the moving direction and driving behaviors of floating car at low speed are dynamic and quite complex, disregarded these influence factors will seriously affect the accuracy of the evaluation of the Average Link Travel Time. In this paper, we proposed an effective Average Link Travel Time evolution method via combining trajectories with a new road network structure. Finally, we carried out our experiments on real data of 15,000 taxies for 13 months in several ways.