Titulo Estágio
DEXTeR: Discovering EXplicit and ImpliciT Service Relationships in the Web
Área Tecnológica
Sistemas de Informação
Local do Estágio
DEI-FCTUC
Enquadramento
The last decade has seen an increased interest in the study of networks in many fields of science. Examples are numerous, from the Internet, the World Wide Web, social networks, Linked Data, sociology, biology, to physical systems such as power grids. Nonetheless, the field of service networks has received less attention. Previous research has mainly tackled the modeling of single service systems and service compositions, often focusing only on studying temporal relationships between services (e.g. business process composition and software-based service composition). Understanding how services operate as part of a large scale global network, the related risks and gains of different network structures and their dynamics is becoming increasingly critical for society.
To address the growing importance of connecting service systems, we have introduced the concept of Open Semantic Service Networks (OSSN) [5]. A service network is defined as a graph structure composed of service systems which are nodes connected by one or more specific types of service relationship, the edges. A service system is a functional unit with a boundary through which interactions occur with the environment, and, especially, with other service systems. Relationships take the form of associations, membership, dependencies, roles, or financial exchanges among numerous other aspects of business relations. This is what makes service network analysis fundamentally different from, e.g., social network analysis and other network systems: the nature of relationships.
Up to now, the study of online relationships between services on the web has not been considered. While studies exist that have proposed approaches to model service relationships (e.g. [1, 4]), the online identification and classification of explicit and implicit relationships has not been explored so far. This project tackles precisely this goal: discovering explicit and implicit service relationships in the web. Precious relationships information can be discovered from social networking platforms such as LinkedIn, Plaxo, Web sites, service marketplaces, etc. Explicit and implicit service relationships information is a cornerstone to enable service network analysis [2].
Objetivo
A service network is represented as a graph G=(V,E). Each node in the set V represents a service in the network, and an edge (u,v) in the set E models a certain type of relationship or interaction between the services represented by nodes u and v. Depending on the type of relationship modeled the edges may be directed or undirected. In many domains, the network structure includes relationships that are explicitly declared by service provider and relationships that are implicit and have to be inferred.
In the past, the study of relationships (e.g. from the area of supply chains) was usually carried out as a "field study" on small communities, gathering data through questionnaires, interviews, and other labor-intensive methods. A prominent example is the Travers and Milgram experiment which consisted in sending physical letters within the US and which led to the discovery of the concept of "six degrees of separation". In contrast, in our approach we seek to develop fully automated methods. Web crawling and scraping techniques can be used to analyse unstructured service sources from the Web, such as corporate Web sites and marketplaces (e.g. social networks such as LinkedIn, service marketplaces such as ServiceMagic.com, Sears’ ServiceLive.com, ServiceAlley.com, and Redbeacon.com), and create relationships instances on-the-fly.
This project relies on web data sources to discover implicit and explicit service relationships. Two objectives will be achieved:
Objective 1. To classify and discover explicit relationships betweens services. This objective is challenging since the notion of service has not been up to now clearly defined by providers. While in social networking platforms such as FaceBook, individuals can declare explicitly their "friends", what they "like", "join" a group, "follow" another user, accept a "friendship" request, etc., service web sites and marketplaces have not followed the same direction. Therefore, the first step is to identify which explicit web information can be used to establish explicit relationships betweens services. Naturally, social and service platforms can contribute with valuable information. Service marketplaces such as ServiceMagic.com, Sears’ ServiceLive.com, and ServiceAlley.com contain information classifying services, and indicating relationships between services.
Objective 2. To classify and discover implicit relationships betweens services. However, these explicitly declared relationships identified in Objective 1 may be incomplete and not describe entirely all of the relationships in the network. The missing links are implicit and need to be discovered using appropriate methods. Implicit connections can be discovered from services’s activities by analysing extensive and repeated interactions between service providers and consumers. While in social media sites, this may include voting, sharing, bookmarking, tagging, and/or commenting items from a specific user or set of users, in service platforms this information is not yet clearly defined and studied. For example, we can think of using the similarity of services to establish connections. This is one of many approaches where the same tags are often used to describe similar services. As with objective 1, the same service marketplaces (e.g. ServiceMagic.com) also contain information which can be processed to identify implicit relationships. To improve the precision of this objective external data sources, such as Wikipedia and and FaceBook, need to be considered.
Both objectives will be materialized with a software tool which will crawl the web, and identify explicit and implicit relationships to better understand how service-oriented economies are organizing themselves as networks.
Plano de Trabalhos - Semestre 1
(a) Literature review (Sept-Nov 2013). Conduct a literature review of open service networks, service relationships, and social network analysis. Read the papers published by the ISG group, namely [2, 3, 5, 1, 4]. Look into the Linked USDL (linked-usdl.org) and OSSR specifications (available at genessiz.org) for modelling services and relationships, respectively.
(b) Web data sources (Oct 2013). Analyse web sites, such as Focus.com, LinkedIn, Wikipedia, ServiceMagic.com, Sears’ ServiceLive.com, ServiceAlley.com, and Redbeacon.com, to find valuable sources which contain the required information to identify service relationships which are fundamental to re-construct open service networks.
(c) Use Case (Nov-Dec 2013). Identify a solid business use case definition which will show the importance of service relationships and that will drive the research project until completion. Create a wiki page and logo for the project.
(d) Architecture and tools (Nov-Dec 2013). Design the first draft of the application to developed to identify explciit and implicit service relationships. Identify the tools and the software needed for the project. For example, scrapper, parsers, and social network and visualization packages.
(e) Mid-term report (Dec-Jan 2013/4). Write and defend the mid-term report.
Plano de Trabalhos - Semestre 2
(a) First prototype (Feb-Mar 2014). Generate the first demo version of the application which identifies service relationships and visualizes the results. Show that your prototype is aligned with the business use case definition (created in step 1.c)). The prototype will help defining additional functional requirements and will also allow receiving feedback from the member of the group.
b) Programming (Mar-May 2014). Program the components, modules and application. Define a second version of the business use cases to better illustrate the benefits of service relationship identification. Definition and execution of the acceptance tests. Receive and address feedback.
(c) Algorithms (April 2013). Adapt existing algorithms from the field of social networking to service networks and that consider explicit and implicit service relationships.
(d) Documentation (Feb-Jun 2014). Integration and revision of the documentation produced in each phase of the project into the final documentation which describe the work carried out technically and from a business perspective. Update weekly a wiki with the project information. Write the final version of the business use case.
(e) Thesis (Jun 2014). Writing of the MSc thesis. Take the final master thesis and formatted it with a suitable format to be submit to a conference.
Condições
This work will be carried out at DEI/Universidade de Coimbra within the Information System Group (ISG). A suitable space will be made available to students. Meetings will be held regularly (every week) with several students working on service networks, service systems, and service relationships.
Orientador
Jorge Cardoso/Alexandre Pinto
jcardoso@dei.uc.pt 📩