Ali Tahir

Clustering user trajectories to find patterns for social interaction applications

Sharing of user data has substantially increased over the past few years facilitated by sophisticated Web and mobile applications, including social networks. For instance, users can easily register their trajectories over time based on their daily trips captured with GPS receivers as well as share and relate them with trajectories of other users. Analyzing user trajectories over time can reveal habits and preferences. This information can be used to recommend content to single users or to group users together based on similar trajectories and/or preferences. Recording GPS tracks generates very large amounts of data. Therefore clustering algorithms are required to efficiently analyze such data. In this paper, we focus on investigating ways of efficiently analyzing user trajectories and extracting user preferences from them. We demonstrate an algorithm for clustering user GPS trajectories. In addition, we propose an algorithm to correlate trajectories based on near points between two or more users. The obtained results provided interesting avenues for exploring Location-based Social Network (LBSN) applications.

A comparison of open source geospatial technologies for web mapping

The past decade has witnessed a steady growth of open source software usage in industry and academia, leading to a complex ecosystem of projects. Web and subsequently geographical information systems have become prominent technologies, widely adopted in diverse domains. Within this context, we developed an open source web platform for interoperable GIServices. In order to implement this architecture, 14 projects were selected and analysed, including the client-side libraries and the server-side components. Although other surveys have been conducted in this area, little feedback has been formally obtained from the users and developers concerning their opinion of these tools. A questionnaire was designed to obtain responses from the relevant online communities about a given set of characteristics. This article describes the technologies and reports the results of the survey, providing first-hand information about open source web and geospatial tools.

A web-based visualisation tool for analysing mouse movements to support map personalisation

Information overload is a well known issue across many domains. Due to an increase in the quantity of information associated with geographic data, information overload is now also prevalent in the spatial domain. This makes interacting with maps tedious and difficult, as extracting relevant information becomes laborious. Map personalisation offers a solution to this problem. By implicitly monitoring user behaviour and interaction with maps, common patterns, preferences and interests can be identified. Using this approach, personalised maps can be generated which match user preferences and contribute to resolving information overload in the spatial domain. Traditionally data mining techniques are used to identify preferences however, visual analytics has proven useful in detecting interests and patterns not apparent via data mining. This paper presents a visual analysis tool called VizAnalysisTools, which can be used by developers and analysts to detect patterns in Web map usage among groups of users. The knowledge gained through this visual analysis can be used to strengthen map personalisation techniques.

Interpreting map usage patterns using geovisual analytics and spatio-temporal clustering

Extracting meaningful information from the growing quantity of spatial data is a challenge. The issues are particularly evident with spatio-temporal data describing movement. Such data typically corresponds to movement of humans, animals and machines in the physical environment. This article considers a special form of movement data generated through human–computer interactions with online web maps. As a user interacts with a web map using a mouse as a pointing tool, invisible trajectories are generated. By examining the spatial features on the map where the mouse cursor visits, a users interests and experience can be detected. To analyse this valuable information, we have developed a geovisual analysis tool which provides a rich insight into such user behaviour. The focus of this paper is on a clustering technique which we apply to mouse trajectories to group trajectories with similar behavioural properties. Our experiments reveal that it is possible to identify experienced and novice users of web mapping environments using an incremental clustering approach. The results can be used to provide personalised map interfaces to users and provide appropriate interventions for completing spatial tasks.

Comparing close destination and route-based similarity metrics for the analysis of map user trajectories

Movement is a ubiquitous phenomenon in the physical and virtual world. Analysing movement can reveal interesting trends and patterns. In the Human-Computer Interaction (HCI) domain, eye and mouse movements reveal the interests and intentions of users. By identifying common HCI patterns in the spatial domain, profiles containing the spatial interests of users can be generated. These profiles can be used to address the spatial information overload problem through map personalisation. This paper presents the analysis and findings of a case study of users performing spatial tasks on a campus map. Mouse movement was recorded and analysed as users performed specific spatial tasks. The tasks correspond to the mouse trajectories produced while interacting with the Web map. When multiple users conduct similar and dissimilar spatial tasks, it becomes interesting to observe the behaviour patterns of these users. Clustering and geovisual analysis help to understand large movement datasets such as mouse movements. The knowledge gained through this analysis can be used to strengthen map personalisation techniques. In this work, we apply OPTICS clustering algorithm to a set of map user trajectories. We focus on two similarity measures and compare the results obtained with both when applied to particular saptial tasks carried out by multiple users. In particular, we show how route-based similarity, an advanced distance measure, performs better for spatial tasks involving scanning of the map area.

Identifying specific spatial tasks through clustering and geovisual analysis

As peoples mobility has increased so too has the use of web maps and other geo-technologies for navigational purposes. Daily usage of these technologies, either embedded in smart phones or through advanced Web GIS, involves carrying out specific spatial tasks. Such spatial tasks can be of various types, context and also consider the physical environment in which the task is being performed. By capturing and analyzing users map interactions, common behavior along with interests and dislikes can be identified and used as a basis on which to study map personalization and adaptation. Completing a spatial tasks essentially corresponds to a mouse trajectory represented as a series of mouse cursor positions on a web map. The challenge is to extract meaning from this spatio-temporal dataset. With the synergy of exploratory visualization and data-mining techniques, we present a novel approach to automatically identify and validate a number of spatial tasks forming complex mouse trajectories within a given study area. The task validation justifies trajectory clustering techniques as well as assisting in attaching semantics to the visual interpretation. This research opens further avenues for studying map personalization.

Classification Of Breast Cancer Histology Images Using ALEXNET

Training a deep convolutional neural network from scratch requires massive amount of data and significant computational power. However, to collect a large amount of data in medical field is costly and difficult, but this can be solved by some clever tricks such as mirroring, rotating and fine tuning pre-trained neural networks. In this paper, we fine tune a deep convolutional neural network (ALEXNET) by changing and inserting input layer convolutional layers and fully connected layer. Experimental results show that our method achieves a patch and image-wise accuracy of 75.73% and 81.25% respectively on the validation set and image-wise accuracy of 57% on the ICIAR-2018 breast cancer challenge hidden test set.

Using T-Drive and BerlinMod in Parallel SECONDO for Performance Evaluation of Geospatial Big Data Processing

With the growing volume of geographically referenced data, there is a need to develop new approaches for efficient storage and processing of massive spatial data. The increase in spatial data has been fueled by the growing availability of ubiquitous mobile devices such as smart phones equipped with GPS, and the awareness of the usefulness of spatial data and digital earth projects. Geospatial data typically contain vector (points, lines, polygons) and raster (satellite images) components. Increasingly, these data have a temporal aspect when referring to moving objects or mobility data. While there are several Database Management Systems (DBMS) which provide support for spatial queries, there are few that provide support for both spatial and temporal data processing. This article discusses one such geospatial big data analysis platform, Secondo, and tests the performance of simple and complex spatio-temporal queries on datasets of varying sizes. In particular, the efficiency of using parallel processing is investigated. While processing queries on multiple nodes can improve efficiency, our results indicate that such efficiencies depend on a number of factors such as volume of data, complexity of a query and the availability of nodes.