• No results found

An Overview of ViewS Framework

Chapter 3 Research Context

3.2 An Overview of ViewS Framework

This Section outlines a framework for representing, capturing and analysing user viewpoints, called ViewS (Viewpoint Semantics). The formal viewpoint representation together with terms and notation to be adopted in this thesis are defined firstly.

3.2.1 User Viewpoints Representation

Definition of terms.

Social Space. A social space in ViewS includes not only open social web platforms (e.g. Twitter, Flickr) but can also refer to closed environments (e.g. a company's Wiki system, a Virtual Learning Environment) in which users participate and contribute. For referring to users in a social space, the notation for a set of users and for a single user are used.

User Statements. A user statement is a piece of textual content provided by a user. It is an example of user generated content as part of contribution in a social space. For examples, a statement can depict a user's description about an item in an on-line shop, an opinion about a product or an experience when participating in an event. For referring to user statements, the notation for a set of statements and for a single statement are used.

Domain and Topic. A domain refers to a “specified sphere of activity or knowledge”11 in the world. When a domain is split into finer spheres, these

are referred as topics.

Dimension. A dimension in this work is used to define a characteristic that can be used to describe a domain or topic.

Digital Object. A digital object is a digital resource about a topic for which user statements may be collected in a social space. Examples of digital objects include a forum thread about travelling to Greece, a YouTube video footage about a museum visit, a Flickr picture about a music performance during holidays and more. For referring to digital objects, the notation for a set of digital objects and for a single digital object are used.

A Definition for User Viewpoint.

Considering the definitions of terms listed above, a user viewpoint is defined as a tuple:

is a set of users; then the discussion concerns a group

viewpoint;

 is a set of digital objects;

 is a set of statements made by the user(s), and

is a set of ontologies that represent one or more dimensions related

to a domain or topic, or the domain itself;

constitute the input for the viewpoints modelling process. The other two elements ( ) constitute the output as following:

 is a set of ontology entities annotated in , , representing

the semantics of user viewpoints linking to . Hereafter these entities will be called annotated ontology entities.

 is a semantic representation of the user viewpoint focus. The focus

is a semantic projection (overlay) of the annotated ontology entities on the ontology space where:

Figure 3.1 depicts and entity-relation diagram for the viewpoint representation.

11 Oxford Dictionaries Online entry for "domain":

Figure 3.1 An entity-relation diagram describing the relationships between

the viewpoint constituent concepts.

The outline of ViewS is presented in Figure 3.2. The collected UGC, which concerns textual user statements on digital objects, is first pre-processed and then semantically augmented (Component A). The semantic output, in turn, is used for capturing the viewpoint focus (Component B). Components A and B capture the user viewpoints which are then used for analysis and comparison. Each phase is detailed in the following Sections.

User Digital Object Statement makes relates to interacts with n n n 1 1 1 Ontology Entity Ontology is linked to 1 n belongs to 1 1 n Focus has 1 n is shaped by 1 n is shaped in 1 n

Figure 3.2 Outline of the ViewS framework.

3.2.2 Content Collection and Pre-processing

ViewS assumes that there is a way to collect UGC from social spaces – either in closed social spaces or by calling appropriate APIs to access Open Social Spaces, e.g. YouTube. The UGC concerns textual user statements on digital objects.

Digital objects and user statements have to be filtered in order to remove irrelevant or noisy content. The extracted content of interest is then converted to appropriate XML format (see Appendix A.3.1) in order to be semantically augmented in the next phase (Component A).

3.2.3 Component A for Viewpoint Capturing: Semantic Augmentation

The semantic augmentation component comprises three phases for semantic text analysis and annotation (details will be discussed in Chapter 4):

(a) text processing which involves traditional Natural Language Processing (NLP) modules executed on the input text from which the text surface form is extracted,

(b) enrichment of the surface form with linguistic and semantic resources to increase the probability for a textual term to be recognised and mapped to ontologies in the following semantic annotation phase; and (c) semantic annotation to link the surface form and the enriched surface

form to ontology entities.

Content Collection and Pre-Processing

users ( ), digital objects( ), user statements ( )

Component A: Semantic Augmentation

Text processing and semantic annotation: input: user statements ( ) and ontologies ( ) output: links to ontology entities ( )

Component B: Viewpoint Focus Modelling

Semantic mapping and representation on the ontology space

User Viewpoints Analysis Query and Comparison

The semantic augmentation component aims at capturing the semantics of user viewpoints by mapping and extending knowledge about the user statements in ontological spaces. The technical novelty of this component is in the integration of relevant tools to achieve this goal.

3.2.4 Component B for Viewpoint Capturing: Viewpoint Focus Modelling

The viewpoint focus modelling component aims at completing the capturing of user viewpoints for the representation by providing an intelligent mechanism to map and structurally represent the focus of a user or a group of users on the ontology space. The technical novelty of the viewpoint focus modelling is to engineer a computational method that can model the concept of focus as perceived by humans into a computer processable form (details will be discussed in Chapter 6).

3.2.5 User Viewpoints Analysis

The analysis of user viewpoints is directly dependent on the representation and includes querying and comparing user viewpoint models. The novelty of the analysis is to characterise and query viewpoints based on the focus coverage and complexity, as well as to qualitatively compare focus models to explicate similarities and differences.

The query and comparison are enabled with an implemented tool - ViewS Microscope (presented in Chapter 6), which has been used to illustrate the analysis made possible by the framework in the domain of experimentation (Chapter 7).