Re: about hupothesis: [Obi-devel] For scrutiny and comment
Bjoern Peters
part of the core terms for the DENRIE branch, and will be discussed at
the workshop. I think your use of them is perfectly in line with what
was discussed in that branch so far. We still need to work on the
definitions though.
- Bjoern
Alan Ruttenberg wrote:
> I'm ccing this note to the information ontology and protocol
> application list to bring in other's thoughts.
>
> On Fri, Nov 14, 2008 at 9:56 AM, Larisa Soldatova <l...@aber.ac.uk> wrote:
>
>> Alan,
>>
>>
>>> As it happens, statements don't play roles, which are defined to
>>> inhere in independent continuants (i.e. material entities).
>>>
>>>
>> I suppose it depends on how you define statements. If as above (and it is
>> consistent with the ontology of reality), then I agree with you.
>>
>> But if consider a wide sense, i.e a statement encodes(gene, function). It
>> can be a fact, a hypothesis, a result. A simple way to represent it is as
>> roles played by a statement in a certain context.
>>
>
> Well, remembering the general schema for roles, we define a bearer, a
> process in which it is realized, and the mode of participation of the
> bearer in the process. Could you have a go at giving such a definition
> for fact, hypothesis, result? For the sake of argument, let's ignore
> the bearer for the moment, since that has already been identified.
>
>
>>>> I prefer to put it to objectives (and to realizable), not
>>>> to roles, because it is a very important class to us. Roles are not
>>>> supported well by inference engines.
>>>>
>>> Not sure what you mean by this. Could you elaborate?
>>>
>>>
>> Example:
>> robot has_role investigator (in the context of Functional Genomics
>> investigations)
>> robot has_role object of experiment (in the context of AI investigations)
>>
>> Roles are context dependent. But if the context is selected, than the class
>> "robot" should inherit all properties of the class "investigator" or "object
>> of experiment".
>>
>
> What does "selecting a context" mean?
>
>
>> I always prefer to put important classes, which are used a lot for
>> reasoning, not under roles (if it is possible) to simplify the reasoning and
>> to avoid possible errors.
>>
>> Currently, we define hypotheses as informational entities. I try to adjust
>> our representation with OBI and to find a suitable place for them. I do not
>> like it under roles for the above reason. But I can see your point that only
>> "to test a hypothesis", not a hypothesis itself should be an objective. It
>> is correct.
>>
>
> OK.
>
>
>> Can "hypothesis" be a sub class of "information artifact" then?
>>
>
> That makes sense. However, we need a definition that fits. In the past
> we've had hypothesis be a narrative object and defined as being
> labelled ass such. I've been a bit afraid to get more explicit for
> fear of having to explicitly represent quoted things. It seems rather
> complicated.
>
>
>>>> - now RS designs experiments, plate layouts, selects protocols, etc., and
>>>> everything is hypotheses-driven.
>>>>
>>> This process would result in a plan specification in OBI.
>>>
>>>
>>>
>>>> - Then RS executes these experiments, repeats each of them 24 times.
>>>>
>>> The fact that there are 24 repeats means that the plan specification
>>> is in part a replicate design (OBI_0500018).
>>> The execution of the experiment would be an instance of planned
>>> process (OBI_0600062)
>>>
>>>
>> I will need to adjust it all. Your comments are very helpful. Thank you.
>>
>>
>>>> - the results and conclusions are again in the terms of the hypotheses:
>>>>
>>>> investigation has_output result = not affects_growth(C00449,
>>>> delta_YER152C)
>>>>
>>> This is a little trickier if we want to explicitly represent the
>>> conclusion. This sort of fact is what is included in the other OBO
>>> ontologies such as GO, but with one more level of quotation around it.
>>> We haven't explored this yet.
>>>
>> The classes "observation", "result", "conclusion" should be in OBI. Can I
>> propose to consider them for the inclusion? Or it is too late on this stage?
>>
>
> It is never too late. I'm not sure they will be obi versus IAO, but
> that's not central. What is central is having decent definitions. Can
> you offer some?
>
>
>>> I agree with the sentiment. Do you see what I mean about Objective:
>>> verifying hypothesis, versus hypothesis?
>>>
>> Yes, and I agree with you.
>>
>
> Great.
>
>> Larisa
>>
>>
--
Bjoern Peters
Assistant Member
La Jolla Institute for Allergy and Immunology
9420 Athena Circle
La Jolla, CA 92037, USA
Tel: 858/752-6914
Fax: 858/752-6987
http://www.liai.org/pages/faculty-peters