Blog Assignment- COI

Social Network Analysis

Challenge

       Identification of suitable web software that can perceive instances of conflict of interests among the various authors and reviewers within the field of Computer Science.

Background on the use of SNA

       Conflicts of interests are situations that arise from biased decisions. Given the dynamic nature of man, COIs are caused by various numbers of reasons ranging from any defined human interaction or relationship. Detection of COI in the last decade has become emphasized because of the need to have accurate and equal processes like contract acquisitions. In the field of research, detection of COI is imperative in situations where ethical and legal implications have big potentials to damage the writers or target audience. The challenge presented in the research on COI is depicted in ‘connecting the dot’ literature or applications that are existent in a number of study cases. COI is difficult to detect because of the insufficient amounts of information during research. However, in some cases, there are present explicit and implicit amounts of information on the internet that can be applied in the detection of COI.

Data Required

       The type of data required in the SNA comes from bibliographic literature specific to the field of Computer Science. Another type of data required is (friend of a friend) FOAF documents acquired from sources where relationships are explicit. Important to highlight in the information gathering process is the validation of the source through an observation on the levels of integration. For instance, in a database, sites containing different works of the same author should have high levels of integration between them. It is in this that the most preferred data comes from public social networks. Public sources offer the capacities to address the challenge of integration and representation of data as real-world problems. Moreover, public sources are readily available and accessible. Integration of the UCINET analytic software will allow predictive, qualitative, and quantitative evaluation of the FOAF and DBLP documents. This will be done through the measure of collaboration (closeness) between authors in the same literature field.

Important Network Measures

       Given that the collected information comes from persons that have somewhat fixed relationships in the field of academia, the following measures are significant in the SNA.

Social Distance

       This measure refers to the level of divergence between persons in different societal groups. The measure is not physical but assesses divergence under the notions of race, class, gender, ethnicity, and economics. Through use of UCINET, the application will highlight the degrees of divergence between authors from different subgroups.

Disambiguation Algorithm

       The measure checks on whether a certain FOAF or DBLP database has multiple references that refer to the same entity.

Name Reconciliation Algorithm

       The measure analyzes context connected with a certain individual to identify similarities and differences. The measure is employed after the identification of multiple references from the same entity.

Co-Author Relationship

       The COI measure assesses the weight of relationships between authors to identify collaboration of assumptions and biases in their writings. This is the degree of closeness or togetherness applied in predictive analysis.

Applications of SNA to the Challenge

       The SNA is important in the study because it first diagnoses the problem. Through the identification and measure of co-author relationships, SNA identifies the depth of the problem in terms of the level of bias and assumptions made. If similarities are more, the SNA suggests COI is low, but if differences are more, the SNA reveals COI is high. SNA is equally important because the levels of COI act as evaluators of scientific research. In this, low levels will suggest conformity, meaning the pieces of literature are accurate. In meaning, scientific topics that have high COI instances are less applicable given the levels of bias and assumptions. The literature contexts are inaccurate.

Thanks,
Menglin Hong
MIB