Leveraging police incident data for intelligence-led policing
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
This article investigates what can be learned from the data collected in police incident reports by applying cutting-edge analytic techniques: hashing and clustering, to determine the structure of similarity among incidents (clustering); and social network analysis based on co-presence of individuals at incidents (social network analysis). No particular hypotheses are posited; rather, the aim is exploratory, to see what kind of useful information is implicit in the data that is collected by police across the world as a routine part of their operations. The sample for this study consisted of a year’s worth of data from the Kingston Police in Ontario, Canada, comprised of 188 attributes associated with 46,668 incident records. The clustering results provide empirical evidence that the operations of Kingston Police are free from bias with respect to individuals, crimes, or regions. Issues worth further attention are suggested by the clustering. The social network of co-presence, where edges arise from presence at the same incident, also reveal useful properties, highlighting individuals who interact with police a lot, and revealing the role of non-criminals as connectors. While the concept of co-offending, and associated networks, is well-established (e.g., Morselli, 2014), the findings from this chapter suggest value in developing an analogous theory of co-presence networks. This kind of analysis supports intelligence-led policing by helping to identify possible problem areas or opportunities for crime prevention such as hot spot policing. Limitations to improve exploitation of data which police forces already collect include 1) growing the skill sets of analysts so that they can carry out deeper kinds of analysis; 2) providing data analytics infrastructure to enable this kind of analysis; and 3) the difficulty, especially for senior management, in grasping the potential of inductive approaches to large data.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.012 | 0.018 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it