Skip to content. | Skip to navigation

Informatik 5
Information Systems
Prof. Dr. M. Jarke
Sections
Personal tools
You are here: Home Teaching SS 17 Web Science Seminar

Web Science Seminar

Web Science has become an interdisciplinary study field between computer science, mathematics, sociology, economics, and other disciplines. This seminar researches advanced Web Analytics and Web Engineering topics in Web Science probably leading to master thesis topics for excellent students. Topics include: network evolution models and network dynamics, (overlapping) community detection, recommender systems, adaptation and personalization in Web Environments, the Educational Web, Web Trust & Credibility, Web Protocols, Peer-to-Peer Networking for Web Clients, Web-based Software Development Models, particular Web Development methods like Web Components and many more. Students do not only learn to write and present scientific papers but also to peer review them. Students will be assigned to a supervisor helping the student through all steps like literature research, seminar paper and seminar presentation.

The following topics are offered by our supervisors this year:

  • Can the Web Improve Building Information Modeling? - BIM (Building Information Modeling) describes the digital representation of physical and functional characteristics of a facility. It goes beyond covering only the geometry of a building by also including various other properties (e.g. construction details). These models are used to support decision-making and collaboration between different stakeholders involved in the building and management process of a facility. Currently, most BIM processes are supported by proprietary software which use proprietary data-standards. Nevertheless, with the buildingSMART international organization, there exists a foundation that drives the development of open standards (Industry Foundation Classes – IFC) and development processes. The goal of this seminar is to investigate the role of the Web in these processes, how it already supports collaboration and cooperation in BIM processes. We want to investigate Web-based BIM interfaces (for example BIMserver) with a special focus on research in the domain of Web-based BIM modeling.
  • Centrality Measures in Dynamical Networks - Complex real-world systems can be modelled as networks with nodes as agents and edges as interactions among nodes. Complex networks have some basic properties such as power-law degree distribution, small-worldness, shrinking diameter, motifs and community structures. One issue in almost all networks is that all the components, either nodes, edges or the communities, do not have equal importance through the network. In other words, nodes may have different centrality values based on their positions and connections in the network. In this regard, some of the centrality values are computed locally, such as clustering coefficient; however, others can be calculated based on global properties such as eigenvalue centralities. Among the centrality metrics one can mention PageRank and Hyperlink-Induced Topic Search (HITS) and community-aware versions as widely used methods to rank the nodes. In this manuscript, we intend to perform an extensive study on ranking and centrality approaches and investigate their impacts on collective behavior in dynamical phenomena including consensus, synchronization, cascaded failures and information spread.
  • Conflict-Free Replicated Data Types - Developed in parallel and as an alternative solution to Operational Transformation algorithms, CRDTs are data structures which manage causality and provide the mechanisms for merging structures during concurrent editing, making sure that they do not diverge. Through their distributed nature, they can be used in peer-to-peer settings and can scale well in multiple environments. This paper should investigate the existing CRDT algorithms and offer an exhaustive view on the various approaches, highlighting their advantages and disadvantages.
  • Did Social Media Win the Election? – The Influence of the Filter Bubble, Fake News and Personalization on Opinion Making - With the adoption of social network usage through all age groups, news distribution through these channels plays a more and more important role in opinion making. In the course of the 2016 US election, so-called “Fake News” came into the spotlight. These false reports were distributed through social media platforms to harm the opposing presidential candidate. Together with the effects of the filter bubble and Facebook's possibility to exploit it via the dark post functionality, personalized content – fake or not – has a huge influence on shaping opinions (and streamlining them) of individual communities. In this seminar paper, we want to investigate how fake news are distributed through social networks, who profits from it and how it influences the way we use the Web in the future.
  • Model Driven Web Engineering - Model Driven Web Engineering gains a new momentum with the standardization of HTML5 and the connected Web APIs and protocols. In this seminar paper, the past, the present and the future of model driven Web engineering should be explored by main representatives of research and development. A special focus should be given to developer support methodologies.
  • Modeling Mobile Web Applications – Challenges and Opportunities - Current model driven Web engineering techniques don’t cope well with the new challenges mobile Web application development bear. On the other hand, model driven mobile Web engineering might lead to a more structured development approach, improving conceptual design phases. The aim of this seminar topic is to investigate current trends in model driven mobile Web engineering.
  • Participatory Design for Collaborative Web Development in Online Communities of Practice - The topic investigates methods for involving end users in the development of Web applications, which result from collaborative practices in online communities. Some examples include modeling and model driven approaches, participation via various (near real-time) collaboration aspects, collaborative prototyping, collaborative code editing, etc. The work should provide a comprehensive view on the theory and current methods in the field concerning the above mentioned dimensions and offer an overview on the trends in the field.
  • Percolation, Resiliency and Robustness in Complex Networks - Researchers look for mechanism and properties to reliably design and protect complex networks. Resiliency, percolation and robustness are nuclear features pertaining to microscopic, mesoscopic and microscopic elements, i.e. individuals, (overlapping) communities and the whole network. Perturbations imposed on networks can be either random or targeted (attacks). Researches have well studied this area recently; however, we intend to find research niches and investigate their relation to community structures.
  • Quality in Composite Web-based Information Systems - Modern Web applications depend on a huge stack of components on various levels of the Internet infrastructure. For example, countless combinations of databases, services, APIs, and protocols exist to bring value from the back- to the frontend and vice versa. Recently, peer-to-peer based systems add to this complexity due to advances in algorithms like the Blockchain or easier availability of technologies like WebRTC. However, end users normally perceive Web applications as a whole, thus the reputation of applications can decrease fast if individual parts fail. This seminar paper should present the current state-of-the-art of research on quality measures for data, information, and interfaces of composite Web applications in a holistic way. This includes inter alia: service level agreements for self-managing microservice architectures and quality models for component-based Web applications.
  • Role Discovery and Applications in Networks - Role discovery is an emerging research path in network science in which node roles are identified based on topological and structural similarities. Role discovery and community detection are different while the former looks at nodes with similar topological and structural properties; the latter investigates nodes which are densely connected. Role discovery can have different applications such as transfer learning, searching for similar nodes and node classification. Literature has offered supervised and unsupervised approaches to identify node roles. In this regard, we intend to investigate role discovery approaches and state of the art applications in complex networks in order to figure out the open research trends in this area.
  • Security of Peer-to-Peer Networks - Peer-to-peer networks bear many advantages in comparison to classical client server models. Distributed storage of information, an environment controlled by users who run nodes without a central point of authority and no single point of failure. But these advantages come with a price: to protect peer-to-peer networks against security attacks is no easy task. This seminar topic proposes a survey about the development of peer-to-peer network security threads and mechanisms to prevent them.
  • Signed Graphs and Social Balance - Signed social networks can be well represented by graphs with both positive and negative relationships. Positive and negative links indicate trust (friendship) and distrust (enmity). Several social networks like Epinions, Slashdot and Wikipedia have been recently dedicated to the study of signed graphs. Moreover, other contents in social networks can be mapped to a binary case of positive and negative meanings e.g. tweets possess positive or negative feeling. In this topic, we are interested to further scrutinize signed graphs as well as direct our study toward diffusion, overlapping community detection, ranking and sign prediction in networks with both positive and negative connections.
  • The Internet Is for Videos. Can a Decentralized Web Handle It? - The Hypertext Transfer Protocol (HTTP) is the driver of the Web as we know it today. Being a client/server technology essentially, a browser typically sends requests to a server, which then replies with a response. This leads to a situation that increasingly benefits centralized services of opaque global players with vast hosting capacities who control the world's information flow and majority views. Other problems of today's Web include the high affinity (and possibility) of censorship, redirection of requests through DNS spoofing, and traffic throttling or priorization. To this end, a number of technologies, architectures and protocols have been proposed by the research community and industry recently. Particularly, blockchain technologies have gained enormous traction. Another proposed protocol is the InterPlanetary File System (IPFS) that aims to replace HTTP via a peer-to-peer hypermedia distribution stack. In this seminar thesis, a critical overview should be given about efforts trying to fix the issues of today's Web. We are specifically interested in peer-to-peer technologies.
  • User Identity Provisioning on the Decentralized Web - Shopping websites, banking portals and social networks all require users to login in order to fulfill their services. Usually, users have to register for each of these sites separately, and far too often, users reuse passwords across several websites, which leads to security problems, if one of the sites gets compromised. To enable users to login with the same account across different services, open authentication and authorization standards like OAuth2, and more lately, OpenID Connect have been standardized. In theory, websites implementing OpenID Connect redirect users to a central identity provider, where they authenticate and authorize the 3rd party website. In practice, big players like Google and Facebook still dominate the market by only providing identities without consuming others; see also Facebook's recent "Delegate Recovery" proposal. To foster identity provider competition, the research community and industry are increasingly looking into self-issued proofs of identity that work in distributed settings. Although self-signed certificates and the Pretty Good Privacy (PGP) system exist since the 1990s, only recently approaches like WebID-TLS, Mozilla Persona (shut down) and blockchain-related technologies have revived the discussion. This seminar thesis should give a comprehensive overview of the history and state-of-the-art of self-issued identities and related research in distributed Web architectures.
  • Web Video Analysis - Web Videos are ubiquitous nowadays. In this seminar topics we want to survey analysis techniques for Web videos like mining click streams, community detection algorithms and metadata analysis.
  • What Shall We Do! Real-Time Content Recommendations on the Web - With the growing importance of social networks and the ubiquitous availability of Internet connectivity, data streams become an important information source, together with data resulting from real-time behavior. Recommendation systems based on collaboration data, in real-time, represent an emerging topic in the context of social software and (near) real-time interactions on the Web.
 

Topic Assignments

 
Topic Supervisor Author  
Model Driven Web Engineering Klamma 374740
Web Video Analysis Klamma 334996
Can the Web Improve Building Information Modeling? de Lange
Did Social Media Win the Election? The Influence of the Filter Bubble, Fake News and Personalization on Opinion Making de Lange 331427
Modeling Mobile Web Applications –Challenges and Opportunities de Lange 374753
Security of Peer-To-Peer Networks de Lange
Quality in Composite Web-based Information Infrastructures Koren
The Internet Is for Videos. Can a Decentralized Web Handle It? Koren 342759
User Identity Provisioning on the Decentralized Web Koren 362019
Conflict-Free Replicated Data Types Nicolaescu 290765
Participatory Design for Collaborative Web Development in Online Communities of Practice Nicolaescu
What Shall We Do! Real-Time Content Recommendations on the Web Nicolaescu 362008
Centrality Measures in Dynamical Networks Shahriari 344413
Percolation, Resiliency, and Robustness: a Triangle to Design and Protect Complex Networks Shahriari 374315
Role Discovery and Applications in Networks Shahriari 374328
Signed Graphs and Social Balance Shahriari
Document Actions