Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Topic modeling: an approach for research design

Topic modeling: an approach for research design

The following slide will be presented in the 6th Icemal 28 August 2016.
Abstract:
Research design has been a major problem for early career researcher, especially PhD students. This paper discusses new approach to assist a researcher choosing their research topic based on previous papers/studies. Here we use open source tools (Vosviewer and Zotero reference manager) to visualize research topic under keywords "groundwater Jakarta" from close to 256 papers harvested from Google Scholar and Crossref scientific database. Using this approach we can identify the change of main research issues in groundwater resources from the 70’s to date. We also see the research flow path during the period and determine new interesting subtopics. Based on the result we believe that this technique is applicable for postgraduate students
Keywords—topic modeling, data visualization, reference manager, open source, groundwater

482627ef66f15a078118328915d9d092?s=128

Dasapta Erwin Irawan

August 19, 2016
Tweet

More Decks by Dasapta Erwin Irawan

Other Decks in Education

Transcript

  1. Topic Modeling: An Approach for Research Design v 1.0 Dasapta

    Erwin Irawan Institut Teknologi Bandung 0000-0002-1526-0863 Adhi Priyambodho TU Delft The 6th International Conference on Educational, Management, Administration and Leadership Universitas Pendidikan Indonesia 28 August 2016
  2. ICEMAL 2016 UPI 2 Introduction • Research design has been

    a major problem for early career researcher and postgrad PhD students. • Graphical approach may help to reduce the difficulties.
  3. ICEMAL 2016 UPI 3 Current confusions • why are we

    doing the search and what are we exactly looking for? • what sources should be used? • what is the time span of the literature included in the search? • how to identify and select the papers that are most relevant? • when I have collected "enough" data?
  4. ICEMAL 2016 UPI 4 Current practice • Random searching: –

    Without specific key words, – Without analytical thinking in connecting references, – Without quantitative approach in selecting references.
  5. ICEMAL 2016 UPI 5 Proposed practice • Research landscape visualization

    • Multilevel literature searching – Quick search • Visualization – Specific search • Visualization
  6. ICEMAL 2016 UPI 6 Proposed practice • Using open source

    apps: – Reference manager: Zotero (or Mendeley) – PaperMachines plugin – VosViewer
  7. ICEMAL 2016 UPI 7 Proposed practice • Using open source

    apps: – Reference manager: Zotero (or Mendeley) – PaperMachines plugin – VosViewer
  8. ICEMAL 2016 UPI 8 Methods

  9. ICEMAL 2016 UPI 9 Literature time line 2015 2010 2005

    2000 1995 1990 1990 5 10 15 10 20 25 Sum of publication Year
  10. ICEMAL 2016 UPI 10 Results • Network visualization: – contains

    nodes of papers, showing papers with referencing connections and unconnected papers. • Word cloud: – visualization of most frequently mentioned words or phrases, as in the title, abstract, or through out the body of the paper.
  11. ICEMAL 2016 UPI 11 Network visualization

  12. ICEMAL 2016 UPI 12 Network visualization

  13. ICEMAL 2016 UPI 13 Word cloud

  14. ICEMAL 2016 UPI 14 Word cloud

  15. ICEMAL 2016 UPI 15 Word cloud

  16. ICEMAL 2016 UPI 16 Take home message • Useful in

    preliminary stage • As the basis of literature review • Literature analysis works best with proper visualization
  17. ICEMAL 2016 UPI 17 Thank you • For further communication:

    – Email: dasaptaerwin(@)outlook(.)co(.)id – Twitter: @dasaptaerwin