 |
|
| FriendFeed is the easiest way to share online. Learn more » |
Robert Hoehndorf › Comments
Talk: Metarel: An Ontology to support the inferencing of semantic web relations within biomedical ontologies: Ward Blonde
Meena Kharatmal liked this
metarel capture semantics of relations to allow relational inferencing. Meta-ontology for relations, compatible with RO
- Melanie
metarel is a meta-ontology for relations that exists in OBO, compatible with RO
- Robert Hoehndorf
relational closures (transitivity, reflexivity, compositions) created via SPARUL
- Melanie
ontologies + metarel -> RDF -> using SPARUL to create relations inferences while loading into triplestore, which can then be queried using SPARQL
- Melanie
2 hierarchies: relations + relation type. Classes in the former are instances of the latter.
- Melanie
BioGateway: http://www.semantic-systems-bi...
- Melanie
relational closure x has function y, and y is_a z, then x has_function z
- Melanie
composition x is located in y, and y part of z, then x located in z
- Melanie
Metarel: http://www.semantic-systems-bi...
- Melanie
I like this work. However I worry about the areas in which metarel duplicates expressivity that is already present in OWL, as it represents divergence and poses problems for integrating different resources. E.g. transitivity, subproperty already built in. One possible solution - make sure that both representations are present in the source.
- Alan Ruttenberg
Talk: Automated Annotation-based bio-ontology alignment with structural validation: Cliff Joslyn
need for automated ontology alignment with validation
- Robert Hoehndorf
Panel discussion: Ontology and Publishing. Participants are Colin Batchelor, Larry Hunter, Peter Jackson, Alan Ruttenberg, David Shotton, Jabe Wilson
Alan: Realism means: there is stuff out there and we can describe it.
- Robert Hoehndorf
Talk: Providing a realist perspective on the eyeGene Database System: Werner Ceusters
Providing a realist VIEW on the eyeGene Database System would've been a cooler topic for this talk
- Amina Abdulla
eyeGene is a network for eye-disease-causing mutations
- Robert Hoehndorf
eyeGene DB is a repository of genetype and phentype information of patients with eye disease
- Robert Hoehndorf
"A realist view on reality" distinguishs Ontologies, data collections and information models
- Robert Hoehndorf
"A realist view on reality" seems rather circular.
- Robert Hoehndorf
@robert - contrast "a relativist view of reality"
- Alan Ruttenberg
The software demos during lunch were quiet interesting. I was quiet disappointed that they were stopped midway.
Melanie liked this
We need to figure out how to do these better. Problem (at least) - people busy eating somewhere else. Any suggestions?
- Alan Ruttenberg
Having a dedicated time slot for the demos, independent from lunch or break time would help. The demos started late today as there were only a handful of people in the room back from lunch and hence were cut short.
- Amina Abdulla
It would have been nice to start with beer after 8 and have the software demonstrations over beer.
- Robert Hoehndorf
Talk: Towards and ontological representation of resistance: the case of MRSa: Albert Goldfain
Melanie liked this
Based on disease ontology grant
- Robert Hoehndorf
IDO (Infectious Disease Ontology) is a domain ontology based on BFO, part of OBO Foundry
- Robert Hoehndorf
contains Top-level IDO: core ontology of infectious disease domain
- Robert Hoehndorf
IDO extensions: family of reference ontologies for specific diseases and pathogens
- Robert Hoehndorf
Based on dispositional view of disease: disease course exists, bearer of diseasse, part of the bearer (disorder), physical basis of disease
- Robert Hoehndorf
resistence phenomenon: resistance of an individual to a disease, resistance of a tumor to treatment, resistance of a pathogen to a drug, herd immunity of an organism population to a ???, resistance of bacteria to UV, etc
- Robert Hoehndorf
NCI Thesaurus has definition for resistance: uses "resist" in definition, type of resistance unclear: mechanism, function, activity, process? restricted to organism as bearer
- Robert Hoehndorf
desiderata: BFO compatible, positive/active principle: use "lacks" instead of negation; non-proliferation of relations principle: don't add a trivial relation resistant_to, correct granulatiy, pragmatic concerns: IDO/IDO extension terms should mirror scientific interest in resistance types
- Robert Hoehndorf
Resistance as quality vs. disposition: disposition not always manifested, resistance could be quality (sufficiently low susceptibility)
- Robert Hoehndorf
Disposition in BFO 2.0: Dispositions plus Capabilities
- Robert Hoehndorf
Disposition>Capability>Function
- Melanie
But no distinctions between them but their name :) And that capabilities are "good" dispositions and functions "good" capabilities
- Robert Hoehndorf
I can see the capability/function (ie design intent) but I don't know what is a disposition that is not a capability. (wrt to "good" I think there is agreement to not use that as discriminant, and it will be removed - though I may be wrong)
- Melanie
formnalization: 16 triples + 6 inference rules + 5 derived triples
- Robert Hoehndorf
@robert - I'm with you on lack of differentia between capability/disposition
- Alan Ruttenberg
Talk: SNOMED CT's Ontological Commitment: Stefan Schulz
Melanie liked this
Basic question for ontology developers: what are the instances? What are the entities these instances are dependent on?
- Robert Hoehndorf
This is the same as asking for the ontological commitment
- Robert Hoehndorf
Based on Gruber's definition
- Robert Hoehndorf
Truth-values of subsumption and equivalence statements may be true or false
- Robert Hoehndorf
SNOMED CT is increasingly re-designed to fit ontological design principles
- Robert Hoehndorf
SNOMED CT uses the Description Logic EL
- Robert Hoehndorf
Tonsillectomy planned in SNOMED CT
- Robert Hoehndorf
"Denied tonsillectomy" is a subclass of tonsillectomy
- Robert Hoehndorf
Instance of Tonsollectomy planned implied the existence of an instance of Tonsillectomy
- Robert Hoehndorf
every denied tonsillectomy is a tonsillectomy
- Robert Hoehndorf
Problems: the negation of a process is a spcialization of this process - a plan is defined such as its realization is implied - a proper part of a coumpound is its taxonomic parent
- Melanie
Hypothesis: different, competing ontological commitments influenced by the practise of clinical documentation
- Robert Hoehndorf
These are systematic architectural patterns of SNOMED CT: 50,000 procedures have "denied" subconcepts, hundreds of concepts have properties like "planned", "suspected" or "known absent"
- Melanie
SNOMED CT concepts are instantiated by representational artifacts as contained in an electronic health record
- Robert Hoehndorf
Extension of Tonsillectomy includes extension of "Denied Tonsillectomy" is false
- Robert Hoehndorf
Using alternative interpretation: Extension of Record of Tonsillectomy includes extension of "Record of Denied Tonsillectomy" is true
- Melanie
Use of SNOMED CT as an ontology depends on agreement about its ontological commitment
- Melanie
Question: considered including context in the ontology? Answer: context is not in ontology, belongs to information models
- Robert Hoehndorf
hey - it will all be in IAO too :) <self-advertisement/>
- Melanie
@Melanie: Ithink not in the way this is intended; contextualized ontologies would define their categories within a certain setting, the context, and the meaning of the categories may depend on the context
- Robert Hoehndorf
Talk: OBO Applied to Prostate Cancer: James Overton
Based on SNOMED-CT, DICOM (Digital Imaging and Communications in Medicine), RadLex (Radiology Society of North America's Radiology Lexicon)
- Robert Hoehndorf
Based on FMA, Disease Ontology, Protein Ontology, Gene Ontology, PATO, Units of Measurement Ontology
- Robert Hoehndorf
Brings a number of examples from clinical reports
- Robert Hoehndorf
They managed to map certain terms in their report, but gaps in OBO ontologies.
- Melanie
Finds gaps in OBO: several terms missing
- Robert Hoehndorf
Problem with fiat boundaries: vary by convention and application
- Robert Hoehndorf
Medical procedures are missing in the OBO
- Robert Hoehndorf
but they will be in OBI ;) (traditional disclaimer :) )
- Melanie
manual annotation very labour intensive. Many terms in OBO, many not. Current focus of OBO ontologies is on biomedical research, less on medical practice.
- Melanie
Comment by Barry Smith: The only way you can get more funding is by demonstrating usefulness.
- Amina Abdulla
Chris from Mayo clinic: philosophical question before we get to money: you haven't leveraged existing resources. This is more like annotating.We should consider using and leveraging resources such as international classification of procedure and international classification of oncology before building new ones.
- Amina Abdulla
Talk: A set of ontologies to drive tools forthe control of vector-borne diseases: PantelisTopalis
Melanie liked this
Vectorbase: covers 5 different types of vectors
- Robert Hoehndorf
Maybe it is just me, but what is a vector?
- Robert Hoehndorf
Vector (epidemiology), an organism, often an invertebrate arthropod, that transmits a pathogen from reservoir to host
- Amina Abdulla
OBO-Edit snapshots in the presentation look good
- Amina Abdulla
"Only build an ontology if you can take criticism."
- Robert Hoehndorf
Talk: Towards desiderata for an ontology of diseases for the nnotation of biological datasets: Olivier Bodenreider
Melanie liked this
Motivation: several terminologies/ontologies available for describing disease and phenotype, which one to pick?
- Robert Hoehndorf
As a result, set of criteria required to select the right ontology
- Robert Hoehndorf
"Ontology" in this talk includes terminologies and other representations of vocabularies
- Robert Hoehndorf
Biological ontologies often do not link to human diseases, are not widely used, etc
- Robert Hoehndorf
Use of BioPortal to find ontologies; also use UMLS
- Robert Hoehndorf
Desirable characteristics: OBO Foundry characteristics
- Robert Hoehndorf
Some of these are relevant: no restrictions on use, standard format, existent mapping to clinical terminologies, harmonized with other ontologies, regular maintenance, exhaustiveness, support for reasoning (automated)
- Robert Hoehndorf
Top 4: Disease Ontology, SNOMED CT, NCI Thesaurus, UMLS
- Robert Hoehndorf
Get best scores for different reasons
- Robert Hoehndorf
Consider "legacy ontologies" such as UMLS, because they have high coverage and are mapped to clinical terminologies
- Robert Hoehndorf
OBO FOundry principles might be too rigid for the purpose of selecting disease ontologies
- Robert Hoehndorf
Limitations: coarse criteria, arbitrary weighting scheme
- Robert Hoehndorf
Talk: Development of neural electromagnetic ontologies (NEMO): Gwen Frishkoff
Granularity reaches from Cell to Body
- Robert Hoehndorf
ERP components characterized by time, space and function
- Robert Hoehndorf
Few statistical meta-analyses of ERP research, due to lack of integration of data
- Robert Hoehndorf
These slides make me hungry. Broccoli, yum!
- Robert Hoehndorf
Does anybody know the password for the wireless network? Unfortunately, all I can use is a ssh tunnel, but no web.
order served - in your mail :)
- Melanie
Thank you :) This makes blogging much easier :)
- Robert Hoehndorf
Talk: ChemAxiom - An Ontological Framework for Chemistry in Science: Nico Adams
Melanie liked this
There is an ontology plugin for Word which allows the addition of semantic markup in text documents; the markup comes from the OBO ontologies.
- Robert Hoehndorf
Markup can be put into documents a posteriori using NLP mechanisms
- Robert Hoehndorf
Combined with Linked Open Data; permits publication of semantic chemical data as linked data
- Robert Hoehndorf
"The future belongs to the geeks. nobody else wants it."
- Robert Hoehndorf
My comments to the discussion about BFO.
s.pdf 49 KB
Melanie liked this
Great, we finally convinced you to talk :)
- Melanie
I think I will just fight windmills there, but somebody has to try :)
- Robert Hoehndorf
I was happy about this presentation :) (speaking as windmill)
- Alan Ruttenberg
This is my slight update on the slides.
s.pdf 49 KB
Sorry about that, but I have no USB stick, so I will just access my slides from here, and needed to post the update.
- Robert Hoehndorf
Robert Hoehndorf liked this
mid-level ontology, describing information content entities (ICE), processes dealing with ICE, material bearers and relations
- Melanie
Examples of information bearers: books, CDs, Traffic signs, etc.
- Robert Hoehndorf
Often confusion about information and that which it is about.
- Robert Hoehndorf
Problem for IAO: consistency with realist ontology; in what sense does information exist?
- Robert Hoehndorf
stance of IAO: 1. ICE created by sentient, intentionally 2. Information is defined by what it is at creation 3. ICE are related to other things by being about them -> they are dependent entities
- Melanie
Question from Robert: IAO doesn't describe information created by nature, eg sequences? Point for later discussion - response: we may end up with everything being information.
- Melanie
Bee waggle dances are no information, pheromone communication no information, etc; information depends on sentience, and can only be created through intentionality (which Alan seems to ascribe only to humans)
- Robert Hoehndorf
I'm not very happy about that - I agree it is important to have scope, but bee dancing for example communicate information.
- Melanie
Barry: artifacts are created by homo sapiens
- Melanie
Barry: information is a mass noun, and because of that people are using lots of things with it, we should avoid using information as a word on its own
- Melanie
Information is defined by intention at creation.
- Robert Hoehndorf
Information entities do not stand on their own.
- Robert Hoehndorf
measurement data: measurements are often wrong, some always are. A measuerment of something is not that something (though we need to connect them). Many types of measurements. We start with simple case, scalar measurement datum.
- Melanie
this is backwards -- measurements are always correct. They are the experimental data. It's out models that can be wrong.
- Phil Lord
The afternoon session of the tutorial about BFO will be a discussion session. Everybody with problems with BFO is welcome to give a short presentation.
Melanie liked this
BFO, referent tracking, OBI, IAO... no lunch for the braves :)
- Melanie
I am rather tempted to come up with some slides, but it is a bit of a scary thing to do :)
- Robert Hoehndorf
I have some issues too, I may email you so you can put them in your slides :)
- Melanie
I think you should present your slides and I will just sit and blog the discussion :)
- Robert Hoehndorf
Tutorial: Introduction to Ontology of Biomedical Investigations: Alan Ruttenberg
Allyson Lister and Melanie liked this
An ontology covering many domains, built under BFO, collaboration of 19 groups.
- Robert Hoehndorf
"OBI: we rock" (disclaimer: biased blogger ;) )
- Melanie
Materials made and produced for investigations, research objectives, experimental protocols, roles of people in investigations, processing of data gathered in investigations
- Robert Hoehndorf
materials that are not already covered by ontologies covering "natural" materials
- Melanie
knowledge produced by investigation should be harnessed: putting together information produced in 2 different places may lead to new information -> too much to do by hand, we need computers
- Melanie
Several relations in OBI: has_specified_input/output, is_concretization_for, etc. -- I wonder how this fits with sparsity of relations as mentioned in previous slide.
- Robert Hoehndorf
relations in OBI are coming from RO or being added there - though it is a long process to reach consensus and release them.
- Melanie
@Melanie: So are these relations (like has_specified_input and ...-output) distinguished by axioms, or are they, at the moment, just tags for associations between particulars?
- Robert Hoehndorf
PCR products seem to be roles: some piece of DNA is a PCR product in relation to its process of creation...
- Robert Hoehndorf
Plans are cognitive things
- Robert Hoehndorf
Plan specifications are not cognitive.
- Robert Hoehndorf
@frank: Ah, nice. Do you already have some documentation about the axioms of the OBI relations out?
- Robert Hoehndorf
@Robert roles and functions just don't work at this level -- why is a table not a role of a piece of wood?
- Phil Lord
is_proxy_for: example of the chromium release assay
- Melanie
position on gel is proxy for mass and charge of molecule in a western blot etc
- Melanie
JOIN US! :) http://purl.obolibrary.org/obo...
- Melanie
Tutorial: Introduction to Ontology Introductions: Alan Ruttenberg
Relalist ontologies permit an audit trail from representations to elements of the world which can be checked.
- Robert Hoehndorf
Questions to ask: if there is a class, what are the particulars? What tare the entities dependent on - without what else can't they exist? When do entities com into existence. When do they go out of existence? Is it with respect to a certain perspective that an entity can be referred to?
- Robert Hoehndorf
Open world assumption: assertions we make are not complete; what we don't state, or can not infer from what is states, we have no knowledge of; when all possibilities are enumerated, we explicitly say so
- Robert Hoehndorf
Fewer relations are better: fewer possible queries, reasoning about classes is easier than reasoning about relations
- Robert Hoehndorf
reusing terms is hard, but essential: especially in biomedicine, lots of resources already exist, one doesn't want to re do things if not: we have to redefine terms we need (better done by others), need of work for data integration
- Melanie
(yay, finally, some network! :) )
- Melanie
@Melanie: Welcome back!
- Robert Hoehndorf
Tutorial: From BFO to Information Artifact Ontology, day 1, part 1: Barry Smith
Nico Adams, Peter E Midford, Allyson Lister and Meena Kharatmal liked this
Ontologies are about integration of data in biology, across different species, across levels of granularity, across different perspectives, within and across disciplines.
- Robert Hoehndorf
Administrative ontologies: locally developed for specific purposes - reusability and compatibility not so important; can be secret, can be bought and sold, may bring entities into existence by the ontology itself.
- Robert Hoehndorf
Scientific ontologies: must go further, science is global and public and open, etc.; science is "seamless" - all parts must interoperate.
- Robert Hoehndorf
Reasons for building scientific ontologies: many ways to build terminologies, multiple terminologies will not solve data silo problems, need to constrain terminologies so that they converge.
- Robert Hoehndorf
What is to serve as constraint for building ontologies? 1 Authority; 2. First in field. 3. Best candidate terminology (what means "best"?). 4. Voting (on what basis?). 5. Reality, as revealed, incrementally, by experimentally-based science.
- Robert Hoehndorf
ontologies are for re-usuability of data using standards, integration of domains
- Meena Kharatmal
ontologies should be representation of Reality. i would opt for 5.
- Meena Kharatmal
Just managed to get on network - I think I will entrust Robert to report, considering his nice notes :)
- Melanie
@Meena: there was some discussion about this, I think it will be discussed further. Personally, I think (5) is wrong, because it gets things upside-down: before you can find natural laws, you already need the ontology; and you can state the same natural law using very different conceptualizations. The best example is probably geometry, for which complete axiom systems were developed based on many different ontologies.
- Robert Hoehndorf
BFO 1.0 in 2004, based on Aristotelian ontology. First major revision in 2008, adding generically dependent continuants. BFO 2.0 coming soon: incorporate top-level relation from the OBO RO into BFO 1.1, and ensure that all relations internal to BFO (dependence) are included.
- Robert Hoehndorf
New Relation Ontology coming soon, too; about twice the size of original one.
- Robert Hoehndorf
Kinds of relation: universal-universal, individual-universal, individual-individual.
- Robert Hoehndorf
Ontological realism: get real ontology right first and then investigate ways in which this real ontology can be translated into computer-usable form later.
- Robert Hoehndorf
Method of ontological realism: find out what the world is like by doing science, talking to other scientists and working continuously with them to ensure that you don't go wrong; create ontologies adequate to this world, not to some simplified model; build representations of entities in the world, not of concepts
- Robert Hoehndorf
ontologies must be able to change if science changes
- Melanie
main distinction between universal and instance
- Robert Hoehndorf
science texts are about universals, diaries are about instances
- Melanie
avoid word concept :)
- Melanie
ontologies are representation of universals in reality
- Meena Kharatmal
BFO wants to be compatible with higher-order universals, but does not use it at the current moment.
- Robert Hoehndorf
Each term in an ontology represents exactly one universal (and should be referred to using singular nouns.
- Robert Hoehndorf
Three levels: entities in reality, both instances and universals; cognitive representations of this reality; publicly accessible concretizations of these cognitive representations
- Robert Hoehndorf
ontolgoy development starts with level 2 = cognitive representations
- Robert Hoehndorf
and can result in level 3 - concretizations of cognitive representations.
- Meena Kharatmal
extension of universal A is the collection of all particulars A: a class is such a collection
- Melanie
defined classes (eg brother of elvis fan) doesn't correpond to any universal
- Melanie
a scientific ontology is a representational artifact which is intended to represent universals and some defined classes.
- Robert Hoehndorf
Both DOLCE and BFO are heavily influenced by Aristotle (although I am not sure if Claudio Masolo or Nicola would agree with that on the DOLCE part).
- Robert Hoehndorf
BFO is a true upper level ontology, no interference with domain ontologies and issues of cognition, no fiction, "small subset of DOLCE but with a clearer treatment of instances, universals, relations and qualities, time"
- Robert Hoehndorf
On a personal note: the last sentence is rather funny, considering that BFO has no axioms for relations, time, qualities, etc. while DOLCE has a very elabnorate axiom system for these...
- Robert Hoehndorf
time problem: distinction between what exists and what occurs: I exist, this arm movement occurs
- Melanie
process entities = occurrents / continuants = things that exist
- Melanie
parthood depends on time: eg molecule can be part of tumor at one point and not an other -> continuants can gain and loose parts and preserver their identity, not true for occurrents
- Melanie
Occurrents are dependent on a continuant
- Robert Hoehndorf
sameas.org is some project about keeping track of identity (???); but apparently it is bad because they get things wrong about the continuant/occurrent distinction
- Robert Hoehndorf
see current discussion on public-lod about the sameAs website and the ramalteon example - http://markmail.org/message...
- Melanie
Slides are at http://icbo.buffalo.edu/present...
- Robert Hoehndorf
Continuants divided into independent and dependent
- Robert Hoehndorf
One-place dependent continuants and relational dependent continuants.
- Robert Hoehndorf
Dependence relations between processes and substances; also one-place (getting warmer) and relational.
- Robert Hoehndorf
Three dichotomies: instance vs. universal, continuant vs. occurrent, dependent vs. independent
- Robert Hoehndorf
Every functioning (process) is dependent on some function (dependent continuant)
- Robert Hoehndorf
Universal-Universal relations are always All-Some Relations (R(C,D) <-> C subclassof (some R D)
- Robert Hoehndorf
There are some some-some relations. They should not be included in an ontology, because they do not support reasoning. Rather, relevant instance-level relations should be included, as well as probabilistic relations at the type level.
- Robert Hoehndorf
Short interruption about absent nipples and possible fevers.
- Robert Hoehndorf
Relations between occurrents and continuants via relations: dependence, participation, realization.
- Robert Hoehndorf
and that is *why* I didn't post about that example :)
- Melanie
Realization: execution of plan, expression of function, exercise role,...
- Robert Hoehndorf
Realizable dependent entitites: plan, function, role, disposition, algorithm,...
- Robert Hoehndorf
A role is an individual, which exists only while the role is being exercised.
- Robert Hoehndorf
What is a role: realizable entity, which exists because the bearer is in some special physical social or institutional set of circumstances in which the bearer does not have to be, and is not such that, if it ceases to exists, then the physical make-up of the bearer is thereby changed.
- Robert Hoehndorf
student_role instance starts existing at 9.00 am and end up at 12.30; there is being a student_role in this class, and a student_role in general
- Melanie
Disposition (internally grounded realizable entity): a realizable entity which if it ceases to exist, then its bearer is physically changed, and whose realization occurs when this bearer is in some special physical circumstances, in virtue of the bearer's physical make-up.
- Robert Hoehndorf
examples: fragility to break, solubility when placed in liquid..
- Melanie
Dispositions cannot be lost or gained without a physical change in the disposition's bearer.
- Robert Hoehndorf
Capability (new realizable): a disposition that exists in virtue of the bearer's physical make-up, and enables the entity in which it inheres to particpate in events of a certain kind.
- Robert Hoehndorf
yay - speaking french is officially barry-approved a good disposition :)
- Melanie
function = capability that exist due to design (evolution, human...)
- Melanie
The capability definition does not add anything to dispositions, IMHO: the first part is just a consequence of the "physical make-up" part of the disposition def, the second is trivially true without further explanation.
- Robert Hoehndorf
Barry: organisms don't have functions.
- Robert Hoehndorf
Parts of organisms have functions.
- Robert Hoehndorf
There such a thing as a "canonical life plan".
- Robert Hoehndorf
An entity has a biological function iff it is part of an organism and has a disposition to act reliably in such a way as to contribute to the organism's realization of the canonical life plan for an organism of that type.
- Robert Hoehndorf
What is a disease? functions are "good" dispositions, relevant to the realization ofthe canonical life plan; diseases are "counterpart bad dispositions"
- Robert Hoehndorf
Disposition divided into: Function and Disease
- Robert Hoehndorf
I agree with Alan and don't like that good/bad distinction - I am not sure how to define those, and in my opinion they are totally subjective and context-dependent
- Melanie
I don't even like "canonical life plans". Where are the instances?
- Robert Hoehndorf
@Frank: I think they want to claim that diseases can be there even when they are not manifest. So there are processes which are disease-manifestestions, while the disease is the disposition for these manifestations.
- Robert Hoehndorf
Information objects: pdfg file, poems, symbols, sequence, molecular structure, etc.
- Robert Hoehndorf
the disease course is the collection of processes
- Melanie
Information objects are generically dependent continuants.
- Robert Hoehndorf
GDCs are divided into information objects and sequences.
- Robert Hoehndorf
i would like to think as: disease is a "state" of body, as "healthy" is a state of body
- Meena Kharatmal
@Meena: but then, what is a state? You mean DOLCE's statives?
- Robert Hoehndorf
don't like either distinction between information entities and sequences. IMO, sequences, structure etc are information entities.
- Melanie
@Melanie: I agree; I am waiting for the distinctive properties to come up in one of the next slides.
- Robert Hoehndorf
I'm waiting for the IAO talk tomorrow to ask about differentiae :)
- Melanie
Information entities exist in a way which makes them dependent on provenance and on proccessors, in a way in which types are not.
- Robert Hoehndorf
If I am unlucky I will have to restructure my talk about sequences after listening to the tutorials today and tomorrow.
- Robert Hoehndorf
Human is not a universal, but a GDC.
- Robert Hoehndorf
Note: In BFO, every universal is generically dependent on its instances.
- Robert Hoehndorf
@Robert: states represented as properties of objects
- Meena Kharatmal
homo sapiens is the class that exemplifies the concretization of the GDC
- Melanie
@Meena: aaaah, states of affairs; I love states of affairs, especially negative ones :)
- Robert Hoehndorf
@meena - states/properties relation is wrong. This would type properties of objects as "representations", which they are not. They are the things that representations represent
- Alan Ruttenberg
am not sure having the GDC brings anything - if I understood correctly, the GDC is the genome quality that makes the human be a human - seems a bit...over-engineered :)
- Melanie
@Melanie: I wonder which universal could not be rewritten as the class of all things exemplifying a GDC.
- Robert Hoehndorf
@melanie - its a dependency counting thing again. On what things does the genome quality depend
- Alan Ruttenberg
when does it go out of existence?
- Alan Ruttenberg
let me put forth the following: disintegration a process of an object nuclear envelope, prior state of nuclear envelope is continuous and post state of nuclear envelope is fragmented as a result of the process of disintegration.
- Meena Kharatmal
@Meena 1-5 are an artificial distinction. What is the experimental evidence for numbers? The are still useful. Besides, the idea that we should get "reality" right first, then worry about computational applications is non-sensical; it's like suggesting we should get physics right first, then worry about a mathematical representation.
- Phil Lord
ICBO social thread: comment here for party locations, meeting places and common dinner sites :)
Nico Adams and Melanie liked this
hmm seems i am going to miss these due to my virtual existence ;)
- Meena Kharatmal
Too bad :) But maybe we can convince people to have the next conference in Mumbai?
- Robert Hoehndorf
vow gr8, i am so excited about the thought of it :)
- Meena Kharatmal
where are you taking us Robert? :)
- Melanie
Hey, Mumbai is great, and you can just easily get a train to Goa :) And I like conferences at nice beach locations...
- Robert Hoehndorf
So, who will be going on the Niagara Falls trip tonight?
- Robert Hoehndorf
Robert Hoehndorf liked this
Referent tracking: what and why, basis of BFO relevant to RT
- Melanie
When did Weiss kill Senator Long? Life of colonel Weiss, at some point he shot Will Long, Bodyguards shouting Weiss immediately, Long to hospital, survived for 40 days, then died.
- Melanie
RT is a paradigm dev in 2005 based on BFO: goal keep track of portions of reality and what is believed and communicated about them - originally to track patient info
- Melanie
translating data into knowledge, RT wants to make digital copy of the world Requirements: faithful, of everything, through history, computable
- Melanie
distinction between ontologies and information models - ontologies represent only what is always true about entities - IM represents only artifacts in which info is recorded
- Melanie
perfect semantic tools useless if data of bad quality: prevailing EHR don't have enough quality level
- Melanie
2 different disorder codes for same patient on same day refer to different fractures or the same? if same code is used for same patient on different dates can they denote the same fracture?
- Melanie
can the same code applied to 2 patients denote the same fracture?
- Melanie
These slides seem to generally have rather low contrast.
- Robert Hoehndorf
not enough referential capacity of generic terms to denote specific entities; goal of RT =be explicit, use UID instead of words. RT uses instances at specific time: codes for types and UID for instances
- Melanie
there was an attempt to dim the lights - unsuccessful :(
- Melanie
importance of temporal indexing: x part_of y at time t
- Melanie
all the bloggers seem to have abandoned the tutorials for the day... well, maybe more tomorrow.
- Robert Hoehndorf
Robert Hoehndorf, Peter E Midford and Allyson Lister liked this
oops - round of introduction :)
- Melanie
Good idea, since there are no name tags yet :)
- Robert Hoehndorf
componenthood: - components are roughly: mostly bona fide parts that are functional units a bona fide object is an object with close boundary, which boundaries correspond with discontinuity of reality fiat don't cause discontinuity, eg left part of table and right part of table
- Melanie
=> not all proper parts of a whole are components (e.g. arch of aorta)
- Melanie
=> containment is not parthood: wallet is not part of backpack
- Melanie
anchors identified by local alignement: by linguistic analysis, they (Bodenreider et al) were ablt to map 4% of FMA and 9% of Galen
- Melanie
And how can you make the distinction between parthood and containment formally? Are there guidelines that help to make the distinction? E.g., is the nucleus part-of a cell? How is this different from the wallet-case?
- Robert Hoehndorf
hello, i am meena from mumbai, india. i too will be attending the icbo09 virtually.
- Meena Kharatmal
Tutorial: events, powers and dispositions, day 2, part 2
4 different views: dispositional and categorical properties are "real", mixed view where properties are both dispositional and categorical, then two views in which there are only dispositional or only categorical properties.
- Robert Hoehndorf
Potentiality must be some kind of quasi-actuality.
- Robert Hoehndorf
Aristotle: if there was no potentiality, there would be no change; we have change; therefore, we have potentiality.
- Robert Hoehndorf
Categoricalism: a categorical property in the presence of a strong law of nature has a disposition.
- Robert Hoehndorf
The speaker seems to be an anti-Humean.
- Robert Hoehndorf
Categorical properties are fundamental; put these into an environment where certain laws obtain
- Robert Hoehndorf
Aboutness: directed: points at something beyond themselves; non-existence: pointed at object; fuzziness: details about the pointed at can be fuzzy.
- Robert Hoehndorf
Dispositional monism: disposition, given a stimulus, brings about manifestation (event); the manifestation is again some arrangement of objects with dispositions.
- Robert Hoehndorf
Swineburne's regress: if there are only dispositions, what is it they ever do/make?
- Robert Hoehndorf
Tutorial: events, powers and dispositions, day 2, part 1, by Randell Dipert.
Peter E Midford liked this
Today's focus is on acts and actions.
- Robert Hoehndorf
Auto-disposition: a disposition in object o brings about a change in object o.
- Robert Hoehndorf
Alter-disposition: a disposition in object o to bring about a change in object o' and o != o'.
- Robert Hoehndorf
Axiom for dispositions: If disposition d1 to EventKind e1 at TimeKind/Region T1 inheres_in Continuant c1 at Time t1, then, for all ProcessualContexts ?p1, if c1 is_in_context ?p1 at t1, then there exists a time ?t2 and event ?e1 such that ?e1 is temproally_located_at ?t2, ?e1 is_instance_of E1 and ?t2 part_of T1.
- Robert Hoehndorf
Upper level of an ontology consists of an is-a hierarchy of basic ontological kinds and basic ontological relations.
- Robert Hoehndorf
Candidate relations for events: part-of, overlaps, etc.; then: participates-in, coming into being of an object, going out of being, change of continant's location, velocity, ...
- Robert Hoehndorf
Mental events: psychology necessary first, as prelude for acts and actions.
- Robert Hoehndorf
"My thinking about flowers yesterday" is an event, it occurred or happend, has beginning and end, is not material, is often caused or can cause, has temporal parts.
- Robert Hoehndorf
There exists an information content entity that inheres in me and represents flowers. Continuants particpate_in this thinking-about-event. Thinking-about is a process taking place in my nervous system; the information content entity was present at least while the thinking-about event occurred.
- Robert Hoehndorf
Believing and valuing has event-like characteristics.
- Robert Hoehndorf
4 place intend-relation: intend(Person, Disposition, Information-content-entity: Objective, Time)
- Robert Hoehndorf
Act: An event that isbrought about, in the proper way, by an intending.
- Robert Hoehndorf
UCore-SL (Semantic Layer): Definition (Act): An event that is performed by some Agent.
- Robert Hoehndorf
Kinds of acts: simple acts (intended, without prior means-end reasoning, eemental motor skill), habitualized non-simple acts (like riding a bicycle, calligraphy), non-habitualized non-simple acts (immediately preceding means-end reasoning brings it about in the proper way).
- Robert Hoehndorf
Can we reduce continuants to events or events to continuants?
- Robert Hoehndorf
Are there processes without continuants participating? E.g., "it rains", "it thunders".
- Robert Hoehndorf
Alternatively, 4-dimensionalism (perdurantism); Whitehead, Quine, Rescher.
- Robert Hoehndorf
Classical mereology - 2 kinds of entities that can be quantified over (== objects) and collections of (needing for summing together collections of objects)
- Melanie
1 mereological primitive: P, binary predicate, Pxy == x is part of y
- Melanie
Definitions PPxy: proper part: object has other parts: x is part of y and x isn't identical to y overlaps: some object is part of both x and y, they share a part
- Melanie
(theorem: if x is part of y, x and y overlap. Eg if foot part of body, foot and body overlap) DSxy: x and y are discrete, x and y do not overlap SUM(x, A): first argument is individual, second argument is a collection of objects piece of chalk is sum of all chalk molecules in, every molecule is part of this piece of chalk, and every other part overlaps at least one of the molecule (in fact several) in this piece of chalk
- Melanie
5 axioms: 1. reflexive: every object is part of itself (otherwise we use proper part) 2. transitive 3. antisymmetric: if x part of y and y is part of x, x and y are identical 4. Strong supplementation principle: if x is not part of y, then x has some part that is discrete from y 5. Universal fusion principle: every non-empty collectio of object has a sum: there is some object that is made up of piece of chalk and piece of empire state building
- Melanie
Some theorems: T1. if everything that overlaps x also overlaps y, then x is part of y T2. if x and y overlap the same thing, x and y are identical T3. every collection of individuals has at most one sum: : if x is the sum of A, and y is the sum of A, then x and y are identical
- Melanie
model for classical mereology: - parthood is interpreted as subset relation - proper parthood would be the proper subset relation - overlap would be set of Si, Sj (which are subset of S) where Si intersection Sj is not empty (2 subsets overlap if they have a non empty intersection) - discrete: relation that holds between subsets of S that have non overlapping parts - summation: union of all subsets in our domain
- Melanie
weaker mereology why do we need a weaker mereology: some non intuitive axioms, e.g. - A5: there is some object which is the sum of my left hand and the empire state building, and that seems counter intuitive to some people - if x sum of A and y is the sum of A then x and y are identical: some people think that is too strong: idea is that we can have two distinct objects composed of same collection of same molecules
- Melanie
example: play-doh, molded in a cup. this clay has been sitting home for over a year, removed it this morning and took it out from her backpack. some people would say there is an other object here than a cup, that was not at her house this morning, and that a lump of clay has been pulled out of backpack and not a cup. it looks like the cup has to be distinct from the clay, the cup came to existence only moments ago.
- Melanie
temporal mereology (disclaimer - I may be mereology saturated by now ;) )
- Melanie
Curious about how part of different aspects of things interact. E.g. part of on object, relation to part of on the regions that contain the objects. Or part of for information entity, compared to part of in the concretizations.
- Alan Ruttenberg
temporal version of classical mereology entities: objects (w,x,y,z), collections of objects (A,B,C) and time (s, t) one mereological primitvie: ternary predicate, takes 2 objects and time as argument: Pxyt: object x is part of object y at time t
- Melanie
PPxyt: x is a proper part of y at t: x is part of y t time t and y is not part of x at t Oxyt: x and y overlap at t DSxyt: x and y are discrete at t Ext=: Pxxt (x exists at time t)
- Melanie
@alan: part of information entities are really tough, because the same information entities can be part of another one multiple times; and suddenly, you don't have supplementation anymore, and, arguably, no more part-of relation.
- Robert Hoehndorf
mixture example: lemonade - whenever a portion of lemonade is there, all its molecules have to be there and their sum is the portion of lemonade, but at any other time the lemonade molecules are scattered in other places When lemonade is there, it has to be made up of just these components: we wouldn't have same portion of lemonade if we were having different components, it would be a different lemonade
- Melanie
mereotopologies Cxy: x is connected to y. There is no need for physical attachment, but x is zero distance to y (for a temporal theory we would use a ternary predicate Cxyt: x is connected to y at time t)
- Melanie
Axioms: AC1: connection is reflexive, everything is connected to itself AC2: symmetric, if x connected to y, y is connected to x AC3: if x part of y, then everything that is connected to x is connected to y
- Melanie
the last axioms, AC3, does not seem to be very intuitive: if I have a square y and take a circle in the middle of the square x, then there are things that are connected to x (the remaining parts of y), but it seems weird to say that these parts of y are also connect to y.
- Robert Hoehndorf
@robert: somebody mentioned the same, apparently we are talking about "internal connection", and by definition if you are connected to one, you are connected to the other.
- Melanie
Containment relations addition of vocabulary for material: we add immaterial objects, holes, cavities.... we add a unary region function r which maps an object or region to the region it is located in. r(x) is interpreted as the spatial region at which x is exactly located
- Melanie
AC1: if x is a propert part of y, then x's region is proper part of y's region AC2: x's spatial region is its own spatial region Region containment: RCONxy =: Pr(x)r(y) x is region-contained in y means x's region is part of y's region different of part of for example; my heart is region contained in my middle mediastinal space my larynx is region contained in my neck While the larynx is part of my neck, the heart is not part of the mediastinal space
- Melanie
Convex hull: the convex hull of an object or region is the smallest convex region in which x is region-contained. ch(x) is x's convex hull (convex: between 2 points of the set you can draw a line that stays in your set, eg circles are convex))
- Melanie
Surrounding relation: x is surrounded by y means that x's region is part of y's convex hull while x's region are being discrete from y's regio (x is not located where y is located) example: my pleural space is surrounded by my pleural membrane. A bolus of food is surrounded by the wall of my stomach.
- Melanie
Partial containment: PCONxy =: Or(x)ch(y) x is partially contained in y if x's regions overlap y's convex hull example: my oesophagus is partially contained in my thoracic cavity. My tooth is partailly contained in its socket.
- Melanie
those relations have different properties: RCON is transitive, SUR and PCON are not. example: - my heart is RCON in my mediastinal space, and my mediastinal space is RCON in my thoracic cavity, my heart is RCON in my thoracic cavity - a filling is PCON in my tooth, and my tooth is PCON in its socket, but the filing is not PCON in the socket.
- Melanie
- if my heart is RCON in my mediastinal spac, then any part of my heart is RCON in my mediastinal space - my tooth is PCON in its socket even though some of its parts are not PCON in the socket
- Melanie
X is fragile if and only if whenever X is placed in the right circumstances, it will always cause a process of a certain kind; X participates in this process (in a certain role) and is broken at the end. "Being placed in the right circumstances" is a description of situation types together with a role X must play within these situations.
Well, this will probably not be the best definition ever...
- Robert Hoehndorf
Alternative: X is fragile if some entity similar to X has broken before.
- Robert Hoehndorf
Alternative: The probability of breakage given a stimulus on X is increased.
- Robert Hoehndorf
First - scope of fragile. Are flowers fragile? Is paper fragile? Can a gas be fragile?
- Alan Ruttenberg
Maybe: x is fragile iff there is a possible situation type with a role that x can play, such that whenever x is placed in a situation of this type playing the specified role, a breaking process is caused (through laws of nature).
- Robert Hoehndorf
Alternative: x possesses certain properties and the laws of nature are as they are.
- Robert Hoehndorf
Simple conditional analysis: something x is disposed at time t to give response r to stimulus s, iff, if x were to undergo s at t, x would r.
- Robert Hoehndorf
@alan: good question; I guess, it depends what the outcome of a manifestation of the "fragile" disposition would be; my answer would be that flowers can break, paper can break, gas can break (you can freeze it and then break it); but I can easily see how you would answer that differently.
- Robert Hoehndorf
Reformed conditional analysis of dispositions (based on D. Lewis' paper about finish dispositions): Something x is disposed at time t to r to stimulus s, iff, for some intinsic property b that x has at t and for some time t' after t, if x were to undergo s at t, and retain b until t', s and x's having b would be an x-complete cause of x's giving r.
- Robert Hoehndorf
Tutorial: The ontology of events, powers and dispositions: part 2
The second part will be primarily about dispositions and powers.
- Robert Hoehndorf
Talk will be about physical dispositions.
- Robert Hoehndorf
Examples will be "fragility" and "red" for disposition and non-disposition.
- Robert Hoehndorf
Disposition (power): A dependent continuant whose essence involves a modal or causal nature.
- Robert Hoehndorf
Categorical (quality): non-dispositional property.
- Robert Hoehndorf
Dispositional properties arise in virtue of categoricals (qualities).
- Robert Hoehndorf
Structure + law of nature (modal) -> disposition
- Robert Hoehndorf
Manifestation (realization): The enactment of the disposition in some process or activity (event).
- Robert Hoehndorf
Sometimes we will speak of states of affairs.
- Robert Hoehndorf
There is a difference between sure-fire and statistical dispositions.
- Robert Hoehndorf
Single-track vs. multi-track dispositions: restricted to single type of manifestation vs. multiple types of manifestation.
- Robert Hoehndorf
There is the problem of counting types, when you have sub- and super-types.
- Robert Hoehndorf
Stimulus (conditions/trigger): the (state or) event that combines with a disposition in order for it to be manifested.
- Robert Hoehndorf
Most often, a stimulus event has as its main constituent other dispositions in other objects.
- Robert Hoehndorf
Long discussion about what dispositions can be used for in real-life ontologies;
- Robert Hoehndorf
it seems to be the consensus that dispositions are necessary to understand causal structure of the world
- Robert Hoehndorf
Base (ground): the property in virtue of which a disposistion is had by an object.
- Robert Hoehndorf
All dispositions are dispositions in virtue of some base; losing a disposition means losing this base.
- Robert Hoehndorf
property vs. predicate: fundamental difference between them; predicates are linguistic entities, no ontological entities.
- Robert Hoehndorf
different relations between predicate and property: one -> many, many -> one, none -> many, many -> none
- Robert Hoehndorf
Sparse vs. abundant; sparse properties are those we must have for science, and cannot do without.
- Robert Hoehndorf
Abundant approach: a property corresponds to any unique grouping of objects.
- Robert Hoehndorf
We have a disposition whenever a subjunctive conditional is entailed by the ascription.
- Robert Hoehndorf
However, this definition is rather weak.
- Robert Hoehndorf
Tutorial: The ontology of events, powers and dispositions by Randell Dipert: part 1
Events are ontologically prior to dispositions.
- Robert Hoehndorf
Outline: events and their importance, events in top-level ontologies, defining event and occurrent, kinds of occurrents.
- Robert Hoehndorf
Outline (session 2): mental events, acts, quasi-continuants, relating events to continuants
- Robert Hoehndorf
Survey of ontologies, relating to events: begin with SUMO
- Robert Hoehndorf
SUMO: dual object process, intentional proess, motion, internal change, shape change
- Robert Hoehndorf
SUMO: dual object processes: 2-1, 1-2, 2-2 processes (2 objects beginning, 1 at end, etc.)
- Robert Hoehndorf
DOLCE: perdurants divided into events and statives
- Robert Hoehndorf
UCore 2.0 ontology, created by federal government to unify information of all departments.
- Robert Hoehndorf
UCore: developed by department of justice, national intelligence, homeland security, defense.
- Robert Hoehndorf
UCore: taxonomy for events, rather unprincipled.
- Robert Hoehndorf
UCore: taxonomy for events: exercise event, military event, terrorist event,...
- Robert Hoehndorf
UCore: relations: involvedIn, causeOf, locatedAt, occursAt, distinctFrom, sameAs
- Robert Hoehndorf
BFO: occurrents divided into spatiotemporal regions, temporal, processual entities
- Robert Hoehndorf
BFO: the problem of whether a process boundary is a process already appeared.
- Robert Hoehndorf
BFO 2.0: in occurrents, fiat process part same level as processual entity; processual context (deprecated) same level; instants still part of occurrent tree.
- Robert Hoehndorf
BFO 2.0: specifically dependent continuants in realizable entities, roles, dispositions, dispositions with a subclass "function".
- Robert Hoehndorf
BFO 2.0 is in draft stage, still under discussion
- Robert Hoehndorf
Events may not be spacial /at all/.
- Robert Hoehndorf
Command&Control ontology (own work by R. Dipert): upper, middle, lower level
- Robert Hoehndorf
Barry: events inhere in continuants.
- Robert Hoehndorf
Use Cyc for definitions of terms.
- Robert Hoehndorf
Question about strictly 4D ontologies; Johanna Seibt's GPT with publications to be found at http://person.au.dk/da...
- Robert Hoehndorf
BFO: events occur or happen.
- Robert Hoehndorf
definition in BFO inadequate. Better definition for BFO...
- Robert Hoehndorf
Proposal: it must be possible, that the continuant exists at two different time (instants).
- Robert Hoehndorf
definition of event: x is event, iff there exists temporal propert part of x or x could have a temporal part or x necessarily occupies a temporal instant.
- Robert Hoehndorf
Modal operators in definitions: 1. D. Lewis: there are multiple alternative worlds; 2. There exists an entity, y, such that y is the same narrow natural kind as x, and y has temporal proper parts.
- Robert Hoehndorf
Divisions of events: instant vs. interval, hasParticipant FiatObjectPart vs. hasParticipant FiatObject
- Robert Hoehndorf
repeated vs. continuous (single) events, bona fide vs. fiat (controversial; "process" used for complexes of events, temporally and spatially extended, unified by having common end or beginning or function/purpose)
- Robert Hoehndorf
active vs. stative distinction, activities vs. non-activities
- Robert Hoehndorf
three kinds of change events: going out of existence of one or more continuant, coming into existence, change in one or more inherent attributes (can be treated in terms of coming or leaving existence, therefore 3 can be omitted); fourth is change in location, velocity, acceleration
- Robert Hoehndorf
SIG: Bio-Ontologies: Knowledge in Biology
Melanie liked this
For tomorrow, I'm split between data management and M3 (which has the "metadata" word in it). Which to go to?
- Phil Lord
And we now have power over here in T3 - plenty of sockets on the left-hand side!
- Allyson Lister
Grrr. Nothing in the short-seq SIG
- Oliver Hofmann
Go to http://ff.im/4wOyA for the thread for Robert Hoehndorf's talk on an application of non-monotonic reasoning
- Allyson Lister
Go to http://ff.im/4wQlC for the thread for Eleni Mikroyannidi's talk on abstracting and expanding the FMA
- Allyson Lister
Go to http://ff.im/4wRUM for the tread for Phil Lord's talk on alternate definitions for BFO function and role
- Allyson Lister
Go to http://ff.im/4wTA2 for the thread for Colin Batchelor's talk on SW representation of biochemical event statements
- Allyson Lister
http://ff.im/4wZax for the thread for Michael Dumontier's talk on knowledge representation with structure-based chemical identifiers
- Allyson Lister
http://ff.im/4x1mo for the thread for Clement Jonquet's talk on prototyping a biomedical onto recommender service
- Allyson Lister
http://ff.im/4x30J for the keynote by Barend Mons on wikiprofessional
- Allyson Lister
http://ff.im/4xcjq for the panel session on wikis and collaborative knowledge capture
- Allyson Lister
http://ff.im/4xgmQ for Daniel Faria's talk on GO classes
- Allyson Lister
http://ff.im/4xhmC for Michael Dumontier's talk on RKB
- Allyson Lister
http://ff.im/4xj6c for Matthias Samwald's talk on ontology-based representations of biomedical statements
- Allyson Lister
http://ff.im/4xqyz for Bjoern Peter's talk on the Immune Epitope Database
- Allyson Lister
http://ff.im/4xoIA for Larisa Soldatova's talk on OBI
- Allyson Lister
Will there be a common ontologist's dinner location tonight?
- Robert Hoehndorf
http://ff.im/4xtmz for Allyson Lister's talk on semantic data integration for SBML models
- Allyson Lister
@Robert not sure - are many of us off to the city hall thing this evening? That's where I'm heading... not sure there's anything other than snacks there though...
- Allyson Lister
http://ff.im/4xzrg for Patrice Seyed's talk comparing BFO/DOLCE primitive relations
- Allyson Lister
And that closes the SIG. Thanks to everyone!!
- Allyson Lister
@Frank aww thanks very much!
- Allyson Lister
Other ways to read this feed:
