Some(what) Grand Challenges for Information Retrieval

1. Some(what) Grand Challenges for Information Retrieval - Nicholas J. Belkin A Presentation by Ashley Carter

2. To access this PowerPoint for future reference, please go to the following link: https://webspace.utexas.edu/xythoswfs/webui/_xy-39027938_docstore1-t_kCgbhs5i You will need to enter a password to download the PowerPoint. The password is: 385T

3. “Information-related goals, tasks and intentions” • Characterize and “differentiate among information-related goals, tasks and intentions in some principled manners that go beyond straightforward listings, that will apply across a wide variety of contexts, and from which design principles for IR systems can be inferred.” • Develop “methods for inferring information-related goals, tasks and intentions from implicit sources of evidence, such as previous or concurrent behaviors.” • Go from “characterization and identification of goals, tasks and intentions, to IR techniques which actually respond effectively to them.”

4. “Understanding and supporting information behaviors other than specified search” • People have trouble specifying what it is (that is, what information objects) that would help them to realize their goals. • There is little known about the nature, variety and relationships of different information behaviors, and the situations that lead people to engage in any one behavior, or sequence of behaviors.

5. “Characterizing context” • Identify aspects of context. • We need: • Theories and data which identify contextual features. • Ways to identify appropriate contextual features • IR system techniques which take account of knowledge of these features • Experimental methods and evaluation measures

6. “Taking account of affect” • There is little research about “what affective experiences people undergo during the course of information interaction” • Design IR techniques which can “appropriately [understand and] reduce the occurrence of undesirable affective responses”

7. “Personalization” • We must: • Identify “factors or values in each dimension that would inform personalization on that dimension” • Investigate “how the different factors or types of evidence interact with one another” • Devise “techniques which take account of these results in order to lead to a really personalized experience.”

8. “Integration of IR in the task environment” • “Arrange things such that a person never has to engage in interaction with a separate IR system at all.” • Collaborate with those who construct the tools which support the tasks themselves.

9. “Evaluation paradigms for interactive IR” • Establish some “minimal standards for collection of data in interactive IR experiments, such that the records of the interactions might be cumulated, constituting eventually a kind of “test collection” of IR interactions, on which different analyses and different experiments might be conducted.” • “Such a collection…would still not necessarily address the issue of validity, but it might be an interesting and useful start toward that goal.”

10. “(In)Formal models of interactive IR” • “Norbert Fuhr (2008) has recently published a paper outlining what [Belkin] believe[s] to be the first attempt at developing a truly formal model of interactive IR. • http://thenoisychannel.com/2009/08/10/norbert-fuhrs-probability-ranking-principle-for-interactive-information-retrieval/

11. Conclusion • We must address the interactive nature of IR. • We must “give up the idea of strictly formal models of IR, in favor of realistic and useful models of IR.”

12. To access this PowerPoint for future reference, please go to the following link: https://webspace.utexas.edu/xythoswfs/webui/_xy-39027938_docstore1-t_kCgbhs5i You will need to enter a password to download the PowerPoint. The password is: 385T