Harald Bosch

Spatiotemporal social media analytics for abnormal event detection and examination using seasonal-trend decomposition

Recent advances in technology have enabled social media services to support space-time indexed data, and internet users from all over the world have created a large volume of time-stamped, geo-located data. Such spatiotemporal data has immense value for increasing situational awareness of local events, providing insights for investigations and understanding the extent of incidents, their severity, and consequences, as well as their time-evolving nature. In analyzing social media data, researchers have mainly focused on finding temporal trends according to volume-based importance. Hence, a relatively small volume of relevant messages may easily be obscured by a huge data set indicating normal situations. In this paper, we present a visual analytics approach that provides users with scalable and interactive social media data analysis and visualization including the exploration and examination of abnormal topics and events within various social media data sources, such as Twitter, Flickr and YouTube. In order to find and understand abnormal events, the analyst can first extract major topics from a set of selected messages and rank them probabilistically using Latent Dirichlet Allocation. He can then apply seasonal trend decomposition together with traditional control chart methods to find unusual peaks and outliers within topic time series. Our case studies show that situational awareness can be improved by incorporating the anomaly and trend examination techniques into a highly interactive visual analysis process.

Spatiotemporal anomaly detection through visual analysis of geolocated Twitter messages

Analyzing message streams from social blogging services such as Twitter is a challenging task because of the vast number of documents that are produced daily. At the same time, the availability of geolocated, realtime, and manually created status updates are an invaluable data source for situational awareness scenarios. In this work we present an approach that allows for an interactive analysis of location-based microblog messages in realtime by means of scalable aggregation and geolocated text visualization. For this purpose, we use a novel cluster analysis approach and distinguish between local event reports and global media reaction to detect spatiotemporal anomalies automatically. A workbench allows the scalable visual examination and analysis of messages featuring perspective and semantic layers on a world map representation. Our novel techniques can be used by analysts to classify the presented event candidates and examine them on a global scale.

ScatterBlogs2: Real-Time Monitoring of Microblog Messages through User-Guided Filtering

The number of microblog posts published daily has reached a level that hampers the effective retrieval of relevant messages, and the amount of information conveyed through services such as Twitter is still increasing. Analysts require new methods for monitoring their topic of interest, dealing with the data volume and its dynamic nature. It is of particular importance to provide situational awareness for decision making in time-critical tasks. Current tools for monitoring microblogs typically filter messages based on user-defined keyword queries and metadata restrictions. Used on their own, such methods can have drawbacks with respect to filter accuracy and adaptability to changes in trends and topic structure. We suggest ScatterBlogs2, a new approach to let analysts build task-tailored message filters in an interactive and visual manner based on recorded messages of well-understood previous events. These message filters include supervised classification and query creation backed by the statistical distribution of terms and their co-occurrences. The created filter methods can be orchestrated and adapted afterwards for interactive, visual real-time monitoring and analysis of microblog feeds. We demonstrate the feasibility of our approach for analyzing the Twitter stream in emergency management scenarios.

Visual Classifier Training for Text Document Retrieval

Performing exhaustive searches over a large number of text documents can be tedious, since it is very hard to formulate search queries or define filter criteria that capture an analysts information need adequately. Classification through machine learning has the potential to improve search and filter tasks encompassing either complex or very specific information needs, individually. Unfortunately, analysts who are knowledgeable in their field are typically not machine learning specialists. Most classification methods, however, require a certain expertise regarding their parametrization to achieve good results. Supervised machine learning algorithms, in contrast, rely on labeled data, which can be provided by analysts. However, the effort for labeling can be very high, which shifts the problem from composing complex queries or defining accurate filters to another laborious task, in addition to the need for judging the trained classifiers quality. We therefore compare three approaches for interactive classifier training in a user study. All of the approaches are potential candidates for the integration into a larger retrieval system. They incorporate active learning to various degrees in order to reduce the labeling effort as well as to increase effectiveness. Two of them encompass interactive visualization for letting users explore the status of the classifier in context of the labeled documents, as well as for judging the quality of the classifier in iterative feedback loops. We see our work as a step towards introducing user controlled classification methods in addition to text search and filtering for increasing recall in analytics scenarios involving large corpora.

Thematic Patterns in Georeferenced Tweets through Space-Time Visual Analytics

An exploratory study of the potential of georeferenced Twitter data (using tweets from Seattle-area residents over a two-month period) extracts knowledge about peoples everyday life.

TrajectoryLenses - a set-based filtering and exploration technique for long-term trajectory data

The visual analysis of spatiotemporal movement is a challenging task. There may be millions of routes of different length and shape with different origin and destination, extending over a long time span. Furthermore there can be various correlated attributes depending on the data domain, e.g. engine measurements for mobility data or sensor data for animal tracking. Visualizing such data tends to produce cluttered and incomprehensible images that need to be accompanied by sophisticated filtering methods. We present TrajectoryLenses, an interaction technique that extends the exploration lens metaphor to support complex filter expressions and the analysis of long time periods. Analysts might be interested only in movements that occur in a given time range, traverse a certain region, or end at a given area of interest (AOI). Our lenses can be placed on an interactive map to identify such geospatial AOIs. They can be grouped with set operations to create powerful geospatial queries. For each group of lenses, users can access aggregated data for different attributes like the number of matching movements, covered time, or vehicle performance. We demonstrate the applicability of our technique on a large, real‐world dataset of electric scooter tracks spanning a 2‐year period.

Iterative integration of visual insights during patent search and analysis

Patents are an important economic factor in todays globalized markets. Therefore, the analysis of patent information has become an inevitable task for a variety of interest groups. The retrieval of relevant patent information is an integral part of almost every patent analysis scenario. Unfortunately, the complexity of patent material inhibits a straightforward retrieval of all relevant patent documents and leads to iterative, time-consuming approaches in practice. With ‘PatViz’, a new system for interactive analysis of patent information has been developed to leverage iterative query refinement. PatViz supports users in building complex queries visually and in exploring patent result sets interactively. Thereby, the visual query module introduces an abstraction layer that provides uniform access to different retrieval systems and relieves users of the burden to learn different complex query languages. By establishing an integrated environment it allows for interactive reintegration of insights gained from visual result set exploration into the visual query representation. We expect that the approach we have taken is also suitable to improve iterative query refinement in other Visual Analytics systems.

ScatterBlogs: Geo-spatial document analysis

We presented Scatterblogs, a system for microblog analysis that seamlessly integrates search backend and visual frontend. It provides powerful, automatic algorithms for detecting spatio-temporal ‘anomalies’ within blog entries as well as corresponding visual representations and interaction facilities for inspecting anomalies or exploiting them in further analytic steps. Apart from that, we consider the systems combinatoric facilities for building complex hypotheses from temporal, spatial, and content-related aspects an important feature. This was the key for creating a cross-checked analysis for MC1.

Building an Environmental Information System for Personalized Content Delivery

Citizens are increasingly aware of the influence of environmental and meteorological conditions on the quality of their life. This results in an increasing demand for personalized environmental information, i.e., information that is tailored to citizens’ specific context and background. In this work we describe the development of an environmental information system that addresses this demand in its full complexity. Specifically, we aim at developing a system that supports submission of user generated queries related to environmental conditions. From the technical point of view, the system is tuned to discover reliable data in the web and to process these data in order to convert them into knowledge, which is stored in a dedicated repository. At run time, this information is transferred into an ontology-structured knowledge base, from which then information relevant to the specific user is deduced and communicated in the language of their preference.

Using Large Scale Aggregated Knowledge for Social Media Location Discovery

Geospatial analysis of location-enabled social media networks can be utilized to generate vital insights in areas where situational awareness is important, such as disaster prevention and crisis response. However, several recent approaches struggle under the challenge that only a small fraction of the data is actually provided with precise geo-tags or even GPS information of their origin. In this work we introduce two strategies that are suitable to assign probable locations of origin to social media messages of unknown locations. They are based on aggregated knowledge about the author and/or the textual content of the message. Using our prototype implementation and a collected dataset comprising more than one year of geolocated Twitter data, we evaluate the effectiveness of our strategies. Our results show that we can locate up to 74% of all messages that were written in specific cities and about 20% of messages written in specific districts.