Discovering Spatio-Temporal Co-Occurrence Patterns of Crimes with Uncertain Occurrence Time
Round 1
Reviewer 1 Report
Comments for author File: 
Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
This paper proposes a method to discover spatio-temporal co-occurrence patterns among multiple types of aoristic crimes with uncertain occurrence time and demonstrated the superiority of the proposed method over existing methods. The proposed method is practical and is expected to be used in police strategies and local government policies. I would ask the authors to consider minor revision before acceptance.
1. Figure 3: The labels for the third and fourth graphs should be (c), (d).
2. L. 377: Please add an explanation as to why the candidate patterns have (k+1)-size.
3. Table 2: What does “partial result” mean here? Can it be replaced by participation index (PI)?
4. Figure 7: Can the vertical axis of the graphs be set to participation ratio (PR)?
5. L. 512 “8%”: Where can I find it in the table?
6. L. 517 “when the probability threshold is less than 0.5”: Table 2 only shows that it was prevalent when the probability threshold is 0.1.
7. L. 535 “when the probability threshold is less than 0.4”: Same as above.
8. When generating the probability density function, why not exclude events with longer time span (e.g., more than 12 hours) to reduce the instability of the spatio-time proximity relationship? No need to recalculate, but please let me know the possibilities.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
This is interesting paper that seeks to advance the aoristic analysis method.
On the whole it is well written in terms of methods and analysis – but I have some concerns with terminology and familiarity with discipline/language used in crime and place studies.
1) Please take care with terminology/language. I strongly urge you NOT to use the phrase ‘aoristic crime’. This is not a type of crime. Aoristic analysis is a technique used to estimate the time a crime event when it is unknown (i.e. between two time points). All instances of this should be removed as it is not correct. Aoristic is not a type of crime – it is an analysis tool/method. Please do not add inconsistent terminology to the field.
2) Why are you introducing a further new term – Spatio-temporal co-occurrence patterns. Are these different to crime hot spots or repeat victimisation/near repeat victimisation. There is an established language/literature here. If there is something unique about STCPs please explain usage or use established and appropriate terminology.
3) Why use broken windows theory. The literature on crime concentrations (hot spots) has developed primarily from routine activity theory and crime pattern theory, linked to opportunity. Broken windows is based on crime prevention and not patterns of crime. I also suggest you consider the literature on repeat and near repeat victimisation, and the concept of flags and boosts and consider how this might apply to your method.
4) Why are you discussing co-occurrence – how is it different/similar to repeat and near repeat victimisation. Again, you are introducing a new and unnecessary language. If you are drawing from non-criminology literature then be explicit – but I recommend you then demonstrate how you are applying this to repeat/near repeat victimisation/ or concept of crime linkage
5) I was not clear but does your analysis examine each of the crime types individually (ie assault analysed as separate crime type, commercial burglary analysed as separate crime type) – using your method. This is appropriate considering likely explanations for these offence types but please make this explicit if correct assumption.
6) You need to consider some of limitations – you are using both temporal and spatial uncertainty – which I accept is due to dependencies present in data. However, one of issues with your co-occurrence principle is it assumed crimes that are closer together will be more related – based on opportunities available to offender and lack of guardianship. It is possible that within your data you have two offenders operating in close proximity (but are independent of each other. There are some interesting studies about crime linkage based on spatial proximity but this dependency may be uncertain – and should be flagged as a possible limitation.
7) What might be some of the practical applications of this technique. This could be expanded on and thought though in more detail.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
