Ce diaporama a bien été signalé.
Nous utilisons votre profil LinkedIn et vos données d’activité pour vous proposer des publicités personnalisées et pertinentes. Vous pouvez changer vos préférences de publicités à tout moment.

Divoli & Medelyan: HCIR-2011 Presentation

1 108 vues

Publié le

Presentation at the HCIR-2011 workshop by Anna Divoli (University of Chicago) and Alyona Medelyan (Pingar). Title: Search interface feature evaluation in biosciences

Publié dans : Technologie
  • Soyez le premier à commenter

Divoli & Medelyan: HCIR-2011 Presentation

  1. 1. Search interface feature evaluationin biosciences Alyona Medelyan joint work with Anna Divoli (University of Chicago) HCIR-2011 20.10.2011
  2. 2. Talk Overview Search interfaces and their features Search tasks on the web and in bioscience Experiment description Side-by-side evaluation example Data collected Results
  3. 3. autocomplete facetted refinementresults preview related searches search expansions
  4. 4. Talk Overview Search interfaces and their features Search tasks on the web and in bioscience Experiment description Side-by-side evaluation example Data collected Results
  5. 5. Search Tasks and their Types in Web Search(Kellar et al. 2007) weather exchange rate Fact finding … Information gathering email banking grad schools Browsing travel plans shopping … Transactions … blogs Other news …
  6. 6. Search Tasks and their Types in Bioscience I’d like to find out what kind of animal models of huntington’s disease are out there Fact Finding I need to collect publications by others on connexins & how they relate to our studies Information Gathering I want to find out whether there are any new publications on the mechanism that underlies Golgi cisternal maturation in yeast Browsing
  7. 7. Research Questions and Hypotheses Which search interface features are useful1 for searching the biomedical literature? Hypothesis Users prefer different interface features depending on the search task Which approaches to facetted navigation2 work best for this domain? Hypothesis It’s better to display dynamically computed sets of facets than a complete hierarchical list
  8. 8. Talk Overview Search interfaces and their features Search tasks on the web and in bioscience Experiment description Side-by-side evaluation example Data collected Results
  9. 9. The Study Exploratory short study with 6 bioscientists 2 faculty, 2 postdocs, 2 PhD students Q&A on 3 search types in their work, queries, resources, systems 10-15min in person sessions Main study with 10 bioscientists 2 faculty, 7 postdocs, 1 PhD student Email & 1-2hr in person sessions Side-by-side comparison of anonymysed search interface features Per participant: 1 baseline and 1 own query
  10. 10. Talk Overview Search interfaces and their features Search tasks on the web and in bioscience Experiment description Baseline query Side-by-side evaluation example connexin Data collected Results
  11. 11. 1. AutocompleteA B CD E F GF G
  12. 12. 2. Search ExpansionsA B
  13. 13. 3. Faceted Refinement - linksA B C E D
  14. 14. 3. Faceted Refinement - checkboxesF G I H
  15. 15. 4. Related SearchesA B C DE F G
  16. 16. 5. Search Results PreviewA C
  17. 17. Talk Overview Search interfaces and their features Search tasks on the web and in bioscience Experiment description Side-by-side evaluation example Data collected Results
  18. 18. Talk Overview Search interfaces and their features Search tasks on the web and in bioscience Experiment description Side-by-side evaluation example Data collected Results
  19. 19. Usefulness ratings for interface features & search tasksAutocompletion br Positive ff ig Neutral NegativeQuery expansions br 1 participant ff igFacetted refinement br ff igRelated searches br ff igSearch results preview br ff ig
  20. 20. Summary of Findings,Participants’ Comments Autocomplete is less important: “we feel pigeonholed by suggestions” Facets are useful: “we focus and refine the search all the time” Choose facets wisely: “a large number of facets is overwhelming” Checkboxes are better than links: “we want to select multiple values” Aesthetics are important but what really matters is the content
  21. 21. Conclusions Which search interface features are useful1 for searching the biomedical literature? Facets and results preview are useful for any search task Other features are more useful for browsing Which approaches to facetted navigation2 work best for this domain? Few, query-oriented facets with specific values, in checkboxes! Alyona.Medelyan@pingar.com Anna.Divoli@pingar.com

×