Using community authored content to identify place-specific activities
Bibliographic record
Abstract
Understanding the context of a person's interaction with a place is important to enabling ubiquitous computing applications. The ability for mobile computing to provide information and services that are relevant to a user's current location—which is central to the vision of ubiquitous computing—requires that the technologies be able to characterize the activities that a person may potentially perform in place, whatever this place may be. To support the user as she goes about her day, this ability to characterize the potential activities for a place must support work on a city scale. In this dissertation, we present a method to process place-specific community-authored content (e.g., Yelp.com reviews) to identify a set of the potential activities (articulated as verb-noun pairs) that a person can perform at a specific place and apply this method for places on a city scale. We validate the method by processing the place-specific reviews authored by community members of Yelp.com and show that the majority of the 40 most common verb-noun pairs are true activities that can be performed at the respective place; achieving an average mean precision of up to 79.3% and recall of up to 55.9%. We applied this method by developing a Web-service (the Activity Service) that automatically processes all the places reviewed for a city and provides structured access to the activity data that can be identified for the respective places. To validate that the place and activity data is useful and useable, we developed and evaluated two applications that are supported by the Activity Service: Opportunities Exist and Vocabulary Wallpaper. In addition to these applications, we conducted a design contest to identify other types of applications that can be supported by the Activity Service. Finally, we discuss limitations of the activity data and the Activity Service, and highlight future considerations.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.024 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.015 | 0.006 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.003 | 0.004 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.004 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".