Te Pou - Home

o Te Whakaara Nui

Database of New Zealand mental health research

Print

Status
Completed 1 November 2007

Created
17 September 2009

Last updated
17 September 2009

Searching and Analyzing Social Networking Site in Web 2.0

Investigator(s) / AuthorsTiong-Thye Goh, Y.P. Huang, C.L. Liew

 
Principal contact
Name Dr Tiong-Thye Goh
Email Email address is not available; please contact
keadmin@tepou.co.nz for more information.
The research
Summary We explore the techniques used by other researchers in identifying emotional content in unstructured data, and make use of existing technologies to attempt to identify at-risk bloggers. We test the accuracy of a simple algorithm for scoring the presence of certain key words and phrases in blog entries.
Objectives Blogs provide an outlet for youth to explore and share their emotions with the world. By exploring the possibilities of mining the vast repositories of social networking sites, we hope to be able to identify bloggers who are at risk of suicide so that appropriate intervention can take place.
Study design Using a selection of real blog entries harvested from MySpace.com, supplemented with artificial entries from our research. A sample of approximately 15000 user IDs of members aged 15-24 living in New Zealand of roughly equal gender and age distribution was gathered from MySpace.com.
Methods Quantitative, Feasibility, Pilot
Results An analysis carried out on the top 20 ranked bloggers identified using the wildcard dictionary showed that 25% exhibited signs of depression. On the other hand, a number of false positives also existed, primarily caused by irrelevant content such as quizzes, narrative stories, or biblical passages which contained some key phrases from the dictionaries.
Conclusions The ability to definitively identify bloggers with suicidal tendencies is limited in accuracy. However, the research we have done thus far has been successful in demonstrating that simple techniques are able to identify bloggers who are potentially at risk of suicide. A 14% automated identification rate is certainly a positive step in assisting suicide prevention organisations with better organising their efforts.
Key Descriptors Early Intervention, Management, Stress, Suicide/Attempted Suicide/Self Harm, Promotion
Disciplines Support Work
Settings Community
Diagnostic Categories Depression
Populations General Population
Other Keywords
Ethics approval No
Academic led Yes
Service led No
How were service users involved No involvement
Publication in peer review journal Yes
Supporting information
Uploaded files
Links
  • IEEE conference paper
    http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=4476021&isnumber=4475929?tag=1
 

Page last updated: 7 November 2008