Represent risk #22
Comments
|
@pdez90 @phismith @mark-jensen @cmungall We can focus the discussion on risk here. |
|
Hi Pier, I sent an email actually to the contributors instead of posting on github. On Thu, Sep 24, 2015 at 4:50 PM, Pier Luigi Buttigieg <
Yours Sincerely, |
|
There were no replies to my email- I dont think anyone had a chance to look On Thu, Sep 24, 2015 at 4:52 PM, Priyanka deSouza <[email protected]
Yours Sincerely, |
|
Are we saying that 'risk' would be subclass of bfo:disposition? |
The issue is that risk is a probability that vulnerabilities will be realised. How would you express that? |
|
In past discussions the framing was approximately: |
As risk always exists (whether or not it is realised), treating this as a BFO:disposition is likely to work. Thus the WMO DRR treatment:
We could then think along the lines of:
If probabilities are assigned on the data or information level, these risks-as-dispositions can be filtered by a user so only "probable" ones are represented in a given setting. |
|
From Informed Consent Ontology (http://purl.obolibrary.org/obo/ICO_0000078): From 2009 paper on Biomedical Ethics Ontology (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2725426/): These are aligned with what you are thinking. I think use of ‘or’ in the ICO def is a typo--perhaps they meant “a disposition for…” Also maybe of help: |
|
Looks like we can refine those. Do you think we need a new class or could
|
|
On Thu, Oct 1, 2015 at 12:59 PM, Mark Jensen [email protected]
There is no such thing as a potential future harm To Pier: Talking of realization of risk as an exposure process sounds too narrow. to some other entity BS From 2009 paper on Biomedical Ethics Ontology (
|
|
I'm not sure the status of ICO. Their definition needs a fair amount of On Fri, Oct 2, 2015 at 7:24 AM, Pier Luigi Buttigieg <
|
|
How will the classes created here be related to data outside the ontology - for example, columns in data tables produced for statistical purposes, headings for reports, parameters in models...? |
|
Excellent question. I don’t know. I hope we can acquire examples of data soon. An example I’ve seen in the medical domain: columns containing values (1-4) for “mortality risk”. This data is output from a calculation that takes various clinical information about a patient as input (age, gender, diagnoses, procedures, severity of illness measures, etc). The result categorizes that patent into a quartile: “low, moderate, high, or extreme”. This information is about a patient’s likelihood of dying in the near future with respect to their current clinical picture. These risk assessments are used by hospitals and insurance companies to allocate resources, determine care management protocols, estimate costs, etc. My assumption is that we will need a general term in the ontology for risk with various subclasses, such as ‘risk for death’, ‘risk for flooding’, etc. Do others agree? It has been suggested that risk is a disposition. That entails it be realizable, something that can exist without manifestation, for which a corresponding structural change must occur in a bearer if a disposition is lost. We are currently proposing that vulnerability is also a disposition. In one document I found the following definition: ‘vulnerability’ = a disposition of an entity to change its state in response to some perturbation. It has also been suggested that ‘risk’ be defined with respect to realizing a vulnerability, perhaps in a way that allows quantifying the likelihood of that realization, perhaps by referring to exposure to hazards that precipitate the realization of vulnerabilities. A couple possibilities immediately present: Are there other options? If I bear a ‘vulnerability to sunburn’, do I also bear the like-named risk? Be defining ‘risk’ along the lines of (3), we may avoid such a problem. Following from above: a data point in a “mortality risk” column would be a measure (estimation) of the likelihood of a patient participating in some process resulting in their death. The data is output from a ‘mortality risk calculation process’ and is about a ‘risk of death’ disposition that inheres in the patient. I have a couple concerns, but wait for comments. |
|
Greetings. I attended Pier's presentation at UNEP on Monday and was invited to 'leap in' to SDGIO development so here I am. If my comments are off-track or retracing old ground then I ask in advance that you forgive the newbie. From a UN institutional perspective, one of the major stakeholders in the terminology of risks, hazards and impacts is www.unisdr.org who maintain a terminology for harmonising discussions amongst state agencies at http://www.unisdr.org/we/inform/terminology . I am no domain expert on these matters nor on semantics but it would seem an effort to review their use of these terms (e.g they define risk as a combination of probability and consequence) both as a means for encouraging their contribution to SDGIO (say, helping them to start evolving the vocabulary towards an ontology) but also for starting to show that this is not a UNEP-only exercise. I have similar through ts re WMO and WHO but have not yet checked their status regarding risk-related vocabularies. |
|
From WMO- From ISDR- Comment: This definition closely follows the definition of the ISO/IEC Guide 73. The word “risk” has two distinctive connotations: in popular usage the emphasis is usually placed on the concept of chance or possibility, such as in “the risk of an accident”; whereas in technical settings the emphasis is usually placed on the consequences, in terms of “potential losses” for some particular cause, place and period. It can be noted that people do not necessarily share the same perceptions of the significance and underlying causes of different risks. From ISO- From WHO- |
|
@mickwilson Many thanks for the links and input. These will be very useful in mapping out these semantics. |
I agree. This looks similar to the treatment of exposure in #21, and will be fairly straightforward for users to apply. It also gels with many of the defs added above (e.g. ISO Note 3). In this mode, we can attribute a risk to a hazard and a risk to the consequences of that hazard. For example, risk of flooding and then risk of drowning, risk of property damage, etc. We can link the risks of 'consequences' associated with a given hazard to the risk of the hazard itself. This is a good example of ontological unpacking.
In general, yes, risk links to some sort of negative outcome. Like vulnerability, however, it doesn't have to. Looking at the definitions above, this uncertainty seems typical. I'm unsure on how to resolve this. |
|
I'll create a |
@cmungall I think this will be similar to the medical case, but concern the |
|
Perhaps: risk: The disposition of an entity to encounter one or more hazards and, as a result, to realise one or more of its vulnerabilities. A human's risk of drowning is increased should they be surrounded by water in a water body, which has a hazard disposition. It is even higher should they have nothing to mitigate the hazardous effects of the water (e.g. SCUBA equipment): realising their vulnerability towards drowning is that much closer. The hazard must be declared as well as the vulnerabilities relevant to the entity. For example, a deep lake is a hazard (i.e. has_disposition The realisation of vulnerabilities is a way we can connect to the 'impacts' in the WHO, WMO, ISO, and ISDR definitions. If we have a high concentration of entities which all have a vulnerability towards drowning in some site, then the population-level risk of drowning (@rlwalls2008 an example of summation over individual risks) can be classed as high, modulated by the probability that a corresponding hazard (e.g. a flood) would manifest. Naturally, the probabilities themselves are handled on the data layer. Thus: Where "drowning hazards" can be a defined class filled by inference: any entity that has a "drowning hazard" disposition. @phsmith One thing that doesn't sit too well is having a "disposition to encounter" something. Is that pushing the disposition definition too far? |
|
This new version is much better than the current draft. I agree that usage of 'encountered' is not ideal. It is shorthand for participation in a process---one where hazards are present. I'd modify If I recall correctly, we are working on 'exposure' too? If so, the perhaps modify accordingly: "..participate in some exposure process.." @phismith Thoughts? |
I understand where you're coming from; however, a disposition to participate in a process still doesn't sit well. Perhaps a slight re-phrasing will help: risk = a disposition of a bearer to realise one or more of its vulnerabilities when participating in a process which exposes the bearer to one or more corresponding hazards. This puts less emphasis on the participation (which must still happen) and seems more tuned to things like the "risk of drowning": the risk of realising a
We are and should turn our attention there soon. I think we're settled enough on risk and hazard for the moment. |
|
I think Pier's risk = a disposition of a bearer to realise one or more of its vulnerabilities when participating in a process which exposes the bearer to one or more corresponding hazards. misleadingly places the vulnerabilities in the foreground; I think the risk could be there for two subjects with different vulnerabilities; thus I prefer risk = a disposition of a bearer to participate in a process which exposes the bearer to one or more corresponding hazards. We can infer vulnerabilities in some cases. (But we may have data about risks without having data about vulnerabilities.) |
|
I need to give this more thought (too much to digest at first read), but the Wikipedia articles on risk (https://en.wikipedia.org/wiki/Risk) and risk assessment (https://en.wikipedia.org/wiki/Risk_assessment) provide some useful insight here. We should also look at statistical treatments of risk (some of which is already capture in the comments). |
Yes, the subjects may have different vulnerabilities, but I would think they must have vulnerabilities that correspond to the hazards they are exposed to in order to be 'at risk'. A rock has no risk of drowning as it has no corresponding vulnerability.
here, what the hazards correspond to is not clear (in the previous version, they corresponded to some of the subject's vulnerabilities) |
|
In #pato-ontology/pato#67 (comment) I made a comment about how SDGIO represents risk and vulnerability. In terms of BFO approach, the disposition is:
The vulnerability as well as the risk are not something that are realized through a process. Something is or is not vulnerable, something is or is not under a risk regarding a threat or a danger. So is it a quality ? No because it is not expressed by the material entity itself, it depends of the environment and the attributed value by humans. However, impacted by something is a disposition, which is expressed through a process/event. So, the vulnerability is much more the assessment (iao:data item) of the degree to which the material entity is impacted (sensitivity) due to its intrinsic qualities in the process (classified as hazard event for this material entity) when the material entity and the process share a certain spatiotemporal condition (exposure). This measurement could be complete with the same measurement but with adaptive capacity.
REF:
|
|
The problem with a view along these lines is, I think, that it would imply On Wed, Oct 26, 2016 at 4:08 PM, cfrancois7 [email protected]
|
|
In the approach I proposed, yes, the risk doesn't exist before the assessment is realized. In essence, the risk is the output of the assessment, given in quality (low/medium/high level) or quantify (probability) data. The definition of the risk I consider is the typical definition used in the geography field, and also the definition used by IPCC: a probability/likelihood of a damage realized by a process on material entity. The problem with the vulnerability as a disposition is the vulnerability depends on the duration on which we consider it. For instance, the building where I'm currently writing my lines is sensitive to flooding, but is not vulnerable regarding flooding for the next decades (we are far from the river), but surely will be in fifty years (riverbed is growing). The vulnerability as well as the risk arises with the time just because the exposure arise (spatially) as well as the probability of the event (rising waters) to happen too. A disposition that is function of the probability of the process/event to happen or of the duration of the time window we consider bothers me. A disposition that depends only of condition realized by a process, i.e. be impacted/damaged, much more interesting. The information about the probability to happen is defined in the process description and not in the disposition. Thus, the vulnerability and the risk information is attached (is about) the disposition or the material entity. |
|
On Wed, Oct 26, 2016 at 7:38 PM, cfrancois7 [email protected]
suppose there is a dam, the structure of which is gradually becoming I would say -- being a defender of objective probability -- that the risk is the same; and I think this corresponds to ordinary people's use of BS
|
|
Ok, I'm thinking how to converge my approach and yours. I'll share some comments. |
|
I re-thought my approach about vulnerability and risk. I agree a material entity has the disposition at some time to be vulnerable to something, e.g. be vulnerable to a flooding (exposure x sensitivity to the process). In BFO approach:
On this, I have a question. Has the physical change to be intrinsic only? Or Is an extrinsic change is consider as a physical change? Indeed, the vulnerability change in function of the spatiotemporal region of the material entity. I'm clearly very vulnerable and more vulnerable in the space void or in a volcano regarding hazard event than on my chair but I'm intrinsically the same. However, my point is to change the risk it is enough to mitigate/suppress the trigger causes of the hazard event/process without change physically (intrinsically or extrinsically) your material entity. For instance, with the dam, create a upstream deviation of the river to ease the pressure on the dam. I still argue that the risk is a measurement (qualitative/quantitative) that translates the likelihood of the condition needed to realize this disposition (vulnerability) to happen. I would be glad to discuss on these points above. More basic/standard references for a more common use: |
|
On Thu, Oct 27, 2016 at 12:08 AM, cfrancois7 [email protected]
|
|
Thanks for your paper. |
pbuttigieg
added
the
class released to address issue
Following on from #18 (comment) and #21 (comment)
There is a need to have a class to express "risk" in the SDGIO.
WMO DRR suggests
In the SGIO, this will probably be more a probability of some entity realising an innate vulnerability (disposition) and a mitigated by measures that promote robustness and resilience (pato-ontology/pato#67)
The text was updated successfully, but these errors were encountered: