Politics and Pragmatismin Scientific Ontology Construction<br />Mike TraversInconsistency Robustness 2011<br />
Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />T...
My background<br />SSS<br />Artificial IntelligenceKnowledge RepresentationAgent-based systems<br />Programming Languages<...
Synopsis<br />Knowledge representation inevitably involves inconsistency, controversy, hence politics;<br />Scientific rep...
Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />T...
What’s a knowledge infrastructure?<br />A system of <br />Technologies, <br />Institutions, <br />Standards,<br />and Prac...
Knowledge Infrastructure #1: Science<br />The scientific community<br />An elaborate web of<br />People (scientists and ot...
Knowledge Infrastructure #2: The Semantic Web<br />Set of technical standards for sharing formalized knowledge<br />Aspire...
Provenance<br />Reasoning<br />Classification<br />Relations &Properties<br />Naming<br />
These two are becoming one…<br />Bioscience is by far the largest application area for semantic web technology<br />
Some non-robust properties of the semantic web<br />Too inexpressive(Can’t represent default reasoning or n-way predicates...
Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />T...
Convergence and Controversy<br />Ontologies are supposed to define a common understanding of a domain<br />But “common” is...
Example: psychiatric illness	<br />What constitutes a mental illness?<br />Not at all obvious that categories correspond t...
Psychiatric illness (2)<br />Homosexuality<br />Formerly a pathology, now not, through a highly politicized process<br />A...
Example: category fudging<br />In Pathway Tools, SRI’s bioinformatics knowledge base<br />This is a widely used system for...
Example: Gene/Protein conflation<br />Genes and Proteins are different things<br />But biologists tend to want to use the ...
Gene (DNA)<br />trpA<br />Gene product(Protein)<br />
Search for “trpA”<br />
Moral of this somewhat trivial example<br />There are tensions (inconsistencies) between formal representation and actual ...
Example: how do we classify mitochondria?<br />Organelles (part of cell)<br />But descended from separate endosymbiotic or...
There are consequences<br />“If we accept that mitochondria are bacteria, then the record books have to be rewritten. The ...
Expressivity in Description Logics<br />Description Logics (DL) are the basis for semantic web ontology.<br />Selected lar...
Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />T...
Bruno Latour<br />French philosopher and sociologist of science<br />Roundly reviled for perceived anti-realism<br />Start...
Latour for dummies<br />Science is a social construction (but not an arbitrary one)<br />Network based: a network consists...
Dual face of science<br />Settled science:“That’s the way it is”ObjectiveBlack-boxed<br />Politically Established<br />Nat...
Science in the making:<br />EG: Watson and Crick’s work on the structure of DNA<br />Speculations (A three-strand model wa...
Under construction<br />
Black boxed<br />
Where the representation meets the road<br />Science is: “the transformation of rats and mice into paper”<br />Situated re...
Jeff Shrager, “Diary of an Insane Cell Mechanic”<br />
Intercalation of representations and the phenomenon<br />
Analogizing to KR<br />Knowledge Representation:<br />Realist<br />Objective<br />Settled<br />Factual<br />Established<br...
A new view of the relation between world and representation<br />Latour refocuses epistemology<br />Less on the truth of r...
Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />T...
Realism vs Conceptualism<br />Realism: a movement in philosophy of KR<br />Led mostly by Barry Smith, SUNY Buffalo(eg “Bey...
Realism is extremely annoying<br />Both vacuous and wrong<br />Vacuous: because it presupposes we know what is real before...
Afferent: software for drug discovery chemists<br />
But Realism is Winning	<br />Basis of BFO (Basic Formal Ontology)<br />Which is used by OBO Foundry and other bio-ontology...
Realism only deals with half of science<br />May work for ready-made science, <br />                       hopeless for sc...
Representational Pragmatism<br />Needed: a term with good connotations to compete with “realism”.<br />Connects to a philo...
Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />T...
Some encouraging developments<br />Linked data vs semantic webA somewhat more bottom-up, pragmatic approach to universal k...
BioBike:  a platformfor symbolic biocomputing<br />A web-based, programmable tool for advanced biocomputing<br />Knowledge...
Reworked to be more social<br />Bio-computation<br />Bio-blog menu<br />Knowledge/ data analysis<br />Integration with ser...
Prototype-based KR<br />How the mind categorizes (Rosche, Lakoff)<br />A perennial minority theme in computation:<br />60s...
Biology is prototype-based<br />Every feature of a biological class started out as an exception to a general case!<br />ak...
Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />T...
The Problems<br />Ontologies are plagued with inconsistencies (or compromise) because they are inevitably the product of d...
The Solutions<br />Tools that support how science is actually done, at web scale and with greater visibility and traceabil...
Thank you!	<br />
Upcoming SlideShare
Loading in …5
×

Politics and Pragmatism in Scientific Ontology Construction

1,624 views

Published on

My talk at the Inconsistency Robustness 2011 Workshop...as usual, trying to fit the entire universe into one paper.

Published in: Technology, Education, Spiritual
0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,624
On SlideShare
0
From Embeds
0
Number of Embeds
5
Actions
Shares
0
Downloads
12
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide
  • My glorious backgroundNeeds animating
  • That’s the tower of Babel, an earlier attmept to build a unified language for knowledge sharing…
  • Politics and Pragmatism in Scientific Ontology Construction

    1. 1. Politics and Pragmatismin Scientific Ontology Construction<br />Mike TraversInconsistency Robustness 2011<br />
    2. 2. Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />The nature of actual scientific representation<br />Representational pragmatism<br />Technical directions<br />Conclusions<br />
    3. 3. My background<br />SSS<br />Artificial IntelligenceKnowledge RepresentationAgent-based systems<br />Programming Languages<br />Media Science<br />Human InterfaceConstructionism<br />Visual Programming<br />Scientific Software(@startups, large companies, open source projects, and now SRI)Scientific KMCollaborationDecision Support <br />PublishingStandards<br />Philosophy of Science<br />Sociology<br />Cognitive Science<br />Narrative Theory<br />
    4. 4.
    5. 5. Synopsis<br />Knowledge representation inevitably involves inconsistency, controversy, hence politics;<br />Scientific representation does too, but it has worked-out practices for dealing with it;<br />KR should work more like science rather than the other way around;<br />Representational Pragmatism: a conceptual framework to make it happen<br />
    6. 6. Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />The nature of actual scientific representation<br />Representational pragmatism<br />Technical directions<br />Conclusions<br />
    7. 7. What’s a knowledge infrastructure?<br />A system of <br />Technologies, <br />Institutions, <br />Standards,<br />and Practices <br />that serve to support knowledge <br />Collection<br />Storage<br />Curation<br />Sharing<br />Validation<br />…<br />
    8. 8. Knowledge Infrastructure #1: Science<br />The scientific community<br />An elaborate web of<br />People (scientists and others)<br />Institutions (labs, journals, funding agencies, instrument makers…)<br />Practices (publishing criteria, protocols, conferences)<br />Works pretty well! The gold standard for knowledge in fact.<br />But there are issues of scaling, quality, inertia, siloing, epistemological closure…<br />
    9. 9. Knowledge Infrastructure #2: The Semantic Web<br />Set of technical standards for sharing formalized knowledge<br />Aspires to be a universal framework for knowledge<br />A grand vision of global-scale knowledge representation <br />And tremendously important and needed.<br />
    10. 10. Provenance<br />Reasoning<br />Classification<br />Relations &Properties<br />Naming<br />
    11. 11.
    12. 12. These two are becoming one…<br />Bioscience is by far the largest application area for semantic web technology<br />
    13. 13.
    14. 14. Some non-robust properties of the semantic web<br />Too inexpressive(Can’t represent default reasoning or n-way predicates)<br />Too complex(Prevents widespread acceptance)<br />Too logic-based(Emphasizes wrong things)<br />
    15. 15. Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />The nature of actual scientific representation<br />Representational pragmatism<br />Technical directions<br />Conclusions<br />
    16. 16. Convergence and Controversy<br />Ontologies are supposed to define a common understanding of a domain<br />But “common” is easier said than done<br />In practice:<br />Many different constituencies<br />With different ideas about what’s important<br />Many side-factors complicate things (implementation cost, personal status, existing non-rigorous usages…)<br />Compromise is necessary but rarely produces elegant results<br />
    17. 17. Example: psychiatric illness <br />What constitutes a mental illness?<br />Not at all obvious that categories correspond to real phenomena<br />Huge changes over over time<br />Currently defined by DSM-IV through a highly politicized process<br />History of PTSD (Scott, 1990)<br />“combat fatigue” or cowardice <br />In and out of the DSM<br />Finally recognized as PTSD, partly as response to Vietnam War<br />
    18. 18. Psychiatric illness (2)<br />Homosexuality<br />Formerly a pathology, now not, through a highly politicized process<br />Attention Deficit Disorder<br />Cluster of symptoms, not clear what the boundaries should be<br />Opinions often determined by theories of child-rearing or institutional aspects of school.<br />Insurers and economics are important actors in debate<br />Summary: <br />these disorders are social constructed categories <br />over a definite but unclear underlying reality.<br />
    19. 19. Example: category fudging<br />In Pathway Tools, SRI’s bioinformatics knowledge base<br />This is a widely used system for curating genomes and metabolic pathways<br />Underlying frame system<br />Web based interface<br />
    20. 20.
    21. 21. Example: Gene/Protein conflation<br />Genes and Proteins are different things<br />But biologists tend to want to use the same name for a gene and its product<br />Tension between formal ontology and actual scientific usage<br />Equivalently, an argument between the computer scientists who build the system and the biologists who use it and curate it<br />
    22. 22. Gene (DNA)<br />trpA<br />Gene product(Protein)<br />
    23. 23. Search for “trpA”<br />
    24. 24.
    25. 25. Moral of this somewhat trivial example<br />There are tensions (inconsistencies) between formal representation and actual usage<br />And, software makers end up having to cope with these tensions in design decisions<br />Usually in a kludgy way! <br />Eg, papering over the conflict in the user interface layer<br />Would be nice to have a better theory of how do this.<br />
    26. 26. Example: how do we classify mitochondria?<br />Organelles (part of cell)<br />But descended from separate endosymbiotic organisms <br />With their own DNA<br />(Generally but not universally accepted theory)<br />
    27. 27.
    28. 28. There are consequences<br />“If we accept that mitochondria are bacteria, then the record books have to be rewritten. The first bacterial genome sequence was completed not by American arriviste Craig Venter …in 1995, but instead by … Fred Sanger, who completed the human mitochondrial genome sequence in 1981!”<br />
    29. 29. Expressivity in Description Logics<br />Description Logics (DL) are the basis for semantic web ontology.<br />Selected largely for computational tractability <br />But DL make it hard to do simple things such as representing defaults<br />All cats have hair<br />Except for this one!<br />Expressivity has been traded away<br />A compromise and perhapsnot the right one<br />
    30. 30. Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />The nature of actual scientific representation<br />Representational pragmatism<br />Technical directions<br />Conclusions<br />
    31. 31. Bruno Latour<br />French philosopher and sociologist of science<br />Roundly reviled for perceived anti-realism<br />Started with anthropological studies of science in labs and fields<br />Ends in a rather unique view of representation and even metaphysics<br />
    32. 32. Latour for dummies<br />Science is a social construction (but not an arbitrary one)<br />Network based: a network consists of humans and non-human actors (lab animals, instruments, funding institutions…)<br />Agonistic – trials of strength between networks <br />Understand how science works by tracing the flow of inscriptions, abstractions, and power through these networks<br />An enriched realism, that provides a rich account of the relation between phenomena and representation<br />
    33. 33. Dual face of science<br />Settled science:“That’s the way it is”ObjectiveBlack-boxed<br />Politically Established<br />Natural<br />Science under construction:<br />UnsettledContentiousSearching for allies (people, funding, tools)Building networks of alliance<br />Social<br />
    34. 34. Science in the making:<br />EG: Watson and Crick’s work on the structure of DNA<br />Speculations (A three-strand model was proposed)<br />Contending theories<br />Eventually a winner emerges<br />Science made<br />Now that the structure of DNA is known, <br />it’s a “black box”<br />we can make instruments that measure it<br />representations of its sequence<br />
    35. 35. Under construction<br />
    36. 36. Black boxed<br />
    37. 37. Where the representation meets the road<br />Science is: “the transformation of rats and mice into paper”<br />Situated representations <br />From phenomena<br />Lab notebook<br />Tables in articles<br />Laws of nature<br />Concrete, situated<br />Abstract, objective<br />
    38. 38. Jeff Shrager, “Diary of an Insane Cell Mechanic”<br />
    39. 39.
    40. 40. Intercalation of representations and the phenomenon<br />
    41. 41. Analogizing to KR<br />Knowledge Representation:<br />Realist<br />Objective<br />Settled<br />Factual<br />Established<br />Abstract<br />Graph structures<br />Knowledge Construction:<br />Situated representations<br />Unsettled<br />Bottom-up<br />User interfaces<br />Ad-hoc structures<br />
    42. 42. A new view of the relation between world and representation<br />Latour refocuses epistemology<br />Less on the truth of representations, <br />More on their connection to the world via networks of actants.<br />Should be a natural fit for computationalists<br />Who also make systems of symbols with causal connections to the world and each other<br />
    43. 43. Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />The nature of actual scientific representation<br />Representational pragmatism<br />Technical directions<br />Conclusions<br />
    44. 44. Realism vs Conceptualism<br />Realism: a movement in philosophy of KR<br />Led mostly by Barry Smith, SUNY Buffalo(eg “Beyond Concepts: Ontology as Reality Representation”, 2004)<br />The problem: nobody knows what makes a good ontology<br />His solution: Aristotelian universals<br />Bad ontologies are…those whose general terms lack the relation to corresponding universals in reality, and thereby also to corresponding instances. Good ontologies are reality representations...<br />
    45. 45. Realism is extremely annoying<br />Both vacuous and wrong<br />Vacuous: because it presupposes we know what is real beforehand<br />Wrong: because it doesn’t correspond to actual scientific knowledge representation<br />Examples of failure:<br />Higgs bosons – we don’t know if they are real<br />Genes – were hypothesized before their “implementation” was known; when were they real?<br />Software for synthetic chemistry – mixes real and not-yet-real molecular structures<br />
    46. 46. Afferent: software for drug discovery chemists<br />
    47. 47. But Realism is Winning <br />Basis of BFO (Basic Formal Ontology)<br />Which is used by OBO Foundry and other bio-ontology efforts<br />Nobody wants to be against “realism”… so they picked a good name<br />
    48. 48. Realism only deals with half of science<br />May work for ready-made science, <br /> hopeless for science-in-the-making<br />Where we don’t know what’s real<br />And which is where the action is<br />
    49. 49. Representational Pragmatism<br />Needed: a term with good connotations to compete with “realism”.<br />Connects to a philosophical tradition (James, Peirce, Dewey, Rorty)<br />“It is astonishing how many philosophical disputes collapse into insignificance the moment that you subject them to this simple test of tracing a concrete consequence” -- James<br />Bottom-up rather than top-down; opposed to premature ontologizing; Latourian<br />Support the divergent representational practices of actual science<br />Help science towards convergence, objectivity, and realism, rather than demanding it upfront.<br />
    50. 50. Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />The nature of actual scientific representation<br />Representational pragmatism<br />Technical directions<br />Conclusions<br />
    51. 51. Some encouraging developments<br />Linked data vs semantic webA somewhat more bottom-up, pragmatic approach to universal knowledge infrastructure<br />Freebase, DBPedia similar efforts<br />Open Science movement<br />Open Access Journals (PLoS, etc)<br />Open Data (standards)<br />Open Notebook (practices)<br />
    52. 52.
    53. 53. BioBike: a platformfor symbolic biocomputing<br />A web-based, programmable tool for advanced biocomputing<br />Knowledge-based <br />Programmable<br />Social<br />Really the inspiration of many of the ideas here<br />Joint work with Jeff Shrager (Stanford), Jeff Elhai (VCU), and others<br />
    54. 54.
    55. 55. Reworked to be more social<br />Bio-computation<br />Bio-blog menu<br />Knowledge/ data analysis<br />Integration with services<br />Commentary<br />
    56. 56.
    57. 57. Prototype-based KR<br />How the mind categorizes (Rosche, Lakoff)<br />A perennial minority theme in computation:<br />60s: Sutherland, Sketchpad<br />70s: Early frame-based KR systems<br />80s: Ungar and Smith, SELF programming language<br />90s: Ken Haase, Framer<br />Now: Javascript<br />A structured way to manage inconsistency<br />
    58. 58. Biology is prototype-based<br />Every feature of a biological class started out as an exception to a general case!<br />aka mutation<br />Classes are Aristotelian<br />Prototypes are Darwinian<br />
    59. 59. Overview<br />Introduction<br />Two kinds of knowledge infrastructure<br />Ontological controversies: some examples<br />The nature of actual scientific representation<br />Representational pragmatism<br />Technical directions<br />Conclusions<br />
    60. 60. The Problems<br />Ontologies are plagued with inconsistencies (or compromise) because they are inevitably the product of different interests.<br />Ontologies generally only try to capture the settled science <br />Realism is vacuous, question-begging; if we knew at the start what was real we wouldn't need to do science<br />Knowledge construction is social, tentative, situated, multi-viewpoint, and only objective at its endpoints.<br />
    61. 61. The Solutions<br />Tools that support how science is actually done, at web scale and with greater visibility and traceability<br />A pragmatic view of scientific representation<br />That let scientists work bottom-up from their results<br />that foregrounds the concrete relations between representation and reality (circulating reference)<br />connects science in progress with settled science, supporting and preserving controversy, unsettledness, and argument structure<br />More simply: integrate data and knowledge and the processes that connect them.<br />Open Science: institutions, standards, practices.<br />A representational infrastructure that supports prototypes, default reasoning, and exceptions.<br />
    62. 62. Thank you! <br />

    ×