SIFTER

SIFTER

Research Questions

  1. How do we create an information retrieval system that allows a degree of serendipity, and what balances are important in terms of introducing serendipity?
  2. If we introduce serendipity into personalization, what kind of degradation can we expect, and for how long?
  3. What are the dynamics of introducing serendipity (e.g., loss of precision) ?
  4. What happens to profile dynamics if drift occurs in the document stream?
  5. What happens to profile dynamics if drift occurs in the user profile?
  6. Which observable signals reliably indicate that a user has made an unexpected and useful discovery?
  7. How often are unexpected documents initially judged unhelpful but later recognized as valuable, and what explains that change?
  8. Under what conditions does increasing the variety of search results increase users’ experience of unexpected, useful discovery?
  9. How does a searcher’s prior knowledge influence whether an unexpected document is recognized as useful?
  10. How does feedback exposure affect the accuracy and speed of user-profile estimation?
  11. How does the learning rate influence early concentration on categories and subsequent learning performance?
  12. Under what conditions does diversity in personalized query recommendations lead to discoveries that users experience as both unexpected and valuable? – https://dl.acm.org/doi/10.1145/3770855.3818317
  13. How does the semantic distance between the current document and a recommended document affect user response (evaluation, engagement, and adoption attitude) in the context of next-item news recommendations?  – https://dx.doi.org/10.1145/3808221
  14. How can a search interface be designed to support deferral and subsequent review of serendipitously encountered information while minimizing disruption of the primary search task? – https://dl.acm.org/doi/10.1145/3786304.3788846
  15. Does the inclusion of explicit support for the deferral and subsequent review of serendipitously encountered information change how people handle this information? If so, how? – https://dl.acm.org/doi/10.1145/3786304.3788846
  16. How does the search interface influence subjective and objective performance measures when serendipitous information encounters occur whilst undertaking concurrent search tasks? – https://dl.acm.org/doi/10.1145/3786304.3788846
  17. Can modeling users’ broader, theme-based interests help recommend relevant products from categories they have not previously explored? – https://dl.acm.org/doi/10.1145/3705328.3748133

Annotated Serendipity Datasets

  1. https://github.com/zhefu2/SerenLens – https://dl.acm.org/doi/10.1145/3539618.3591787
  2. https://osf.io/wsakx – https://dl.acm.org/doi/pdf/10.1145/3754459
  3. https://github.com/Ton-moy/surprise-and-serendipity – https://dl.acm.org/doi/pdf/10.1145/3604915.3608851

Timeline

Proposal

  1. Prepare and submit a draft proposal to the Professor: 30 November, 2026.
  2. Submit to the committee: 15 February, 2026.
  3. Tentative defense date: 22 March, 2027 – 1 April, 2027.

Scoping Review

  1. Prepare the Zotero collection. – https://www.zotero.org/groups/6690173/lairhub/collections/3I3EERT4/collection

ACM SIGIR 2027 Paper Preparation

  1. We need to identify the problem statement, methodology, and evaluation.
  2. Last date of submission: January 14, 2027.

Zotero

https://www.zotero.org/groups/6690173/lairhub/collections/H2LVV2U4/collection