The Ontology of CLASSIFICATION Barry Smith ONTOLOGIST c
The Ontology of CLASSIFICATION Barry Smith ONTOLOGIST. c. Om 1
The world knows colour 2
The world knows no redness, greenness, blueness, … These types do not reflect joints in reality 3
The world knows no mild hypertension, moderate hypertension, severe hypertension, … These do not reflect joints in reality 4
The world knows no Poland, Belgium, Utah, Bavaria, … These do not reflect joints in reality 5
The world knows no nation-state, parish, census tract, township, legal jurisdiction … These types do not reflect joints in reality 6
what we think we know what we what the know world knows 7
The world is largely a system of continua, both on the level of instances and on the level of types But what we know (at least as expressed in language) is always in a sense digital rather than analogue 8
All of the mentioned entities arises because of our parcellings, griddings, apportionments, segmentations, . . . = partitions 9
Partitions A partition is the result of drawing a (typically complex) fiat boundary over a certain domain 10
Gr. Gr A simple partition on the level of individuals 11
Gr. Gr 12
Gr. Gr 13
partitions can be extended 14
15
16
partitions can be split and merged 17
partitions can be split and merged 18
partitions can be split and merged 19
A partition can be more or less refined 20
21
22
one partition can be skew to another 23
Perspectivalism Perspectival realism Different partitions which are skew to each other may be equally veridical representations of the same reality 24
We can apply partitions to other partitions 25
Mereotopology of Neuronal Partitions Advanced Database Methodology for the Collation of Connectivity Data on the Macaque Brain Klaas E. Stephan, et al. , Phil. Trans. Royal Society London B, 2001 26
link: Granular partitions http: //ontologist. com 27
An ontology is a partition, or a complex partitions, on the level of types 28
Types and instances Partitions can be created both at the level of instances (Poland vs. Germany) and at the level of types (nation-state vs. colony) 29
An ontology is a partition at the level of types focusing primarily on types (dog, leaf, cell, lung, lake …) whose instances have their own complete bona fide boundaries 30
bona fide objects = objects which exist independently of our partitions (objects with bona fide boundaries) 31
bona fide partition of individuals John Paul George Ringo bona fide partition of types. . . up down charm strange . . . 32
a bona fide type classification 33
But there also fiat partitions 34
Fiat partitions are artefacts of our cognition = of our referring, perceiving, classifying, counting, measuring, mapping activity GRIDDING ACTIVITY 35
Artist’s Grid 36
e. g. they are artefacts for counting 1 2 3 4 37
Frege: “Numbers belong to the realm of concepts” Better: Numbers belong to the realm of partitions 38
Without partitions 39 how many numbers?
Partitions can sometimes create objects fiat objects = objects created by partitions 40
fiat partition at the level of types mild moderate hypertension severe hypertension 41
fiat partition at the level of individuals this tail this torso 42
Kansas 43
Ontology of Maps 44
types represented in an ontology 45
domestic cow breed: HOLSTEIN breed: Brahman species: Bos TAURus 46
A label in the zoo is a mapping between an animal instance and an animal type (all scientific language is built around mappings of this sort between instances and types) 47
California Land Cover 48
a map is a mapping between points California Land Cover in reality and the types represented in a legend x 49
A Map 50
A Portion of Reality 51
A Mapping 52
A Map is a Mapping 53
The Ontology of Measurement 54
55
The world knows no Poland, Belgium, Utah, Bavaria, … These do not reflect joints in reality 56
The world knows no inches, feet, centimeters, seconds, grams, pascals, joules, ohms, teslas, volts, kg/m 3 … These, too, do not reflect joints in reality 57
Measurement belongs to the realm of partitions. . . -20 -10 0 normal 0 10 10 20. . . massively increased chronic . . . 58
59
Partitions can be skew to each other 60
Partitions can be more and less refined and thereby yield measurements which are more or less vague 61
Units of measurement are fiat types Each inch long thing has a length which is an instance of the fiat type inch (the fiat type inch is like the fiat type blue) Each case of mild hypertension has a disorder which is an instance of the fiat type mild hypertension 62
Length is a bona fide type But the different ways of slicing up length (creating a set of quasi-discrete portions out of a continuum) create fiat universals Is there a way of slicing up the universal length into minima? (infinitesimal portions? ) Is Planck length a bona fide unit type? 63
The world knows length continua (as instances) • The world also knows length as a type • which these instances instantiate 64
in the everyday world all the measurements (lengths, temperatures. . . ) with which we have to deal are socially created artefacts subject to phenomena of vagueness, technology-dependence, and using fiat units of measure 65
Phenomenon of vagueness every measurement is vague, to a degree determined by the measuring rod/standard used vagueness is the other side of the coin from granularity every measuring rod (scale) does not recognize differences beneath a certain size 66
Measurement: we lay a grid upon a specific portion of physical reality with a gridding determined by a given unit type, map endpoints to endpoints and count the number of cells between the endpoints 67
An Act of Measurement portion of reality: dependent magnitude (here: distance) + independent bearer 68
The Act of Measurement l l ll l l ll l tape measure (grid) projected onto reality with endpoints mapped to endpoints 69
then you count the cells in the grid 1 2 3 4 70
Ontological assay of the act of measuring length • • act of counting cells in a grid determined by the fiat type which is the pertinent unit of measurement and by a certain portion of reality consisting of some independent continuant bearer together with a dependent feature, here: a certain distance between two endpoints 71
The Periodic Table 72
Ontologies are here 73
or here 74
ontologies represent general structures in reality (leg) 75
Ontologies do not represent concepts in people’s heads 76
They represent universals in reality 77
“leg” is not the name of a concept • • • concepts do not stand in the part_of connectedness causes treats. . . relations used by biomedical ontologies 78
instances A B C 515287 521683 521682 DC 3300 Dust Collector Fan Gilmer Belt Motor Drive Belt universals 79
Inventory vs. Catalog: Two kinds of representational artifact • Databases represent instances • Ontologies represent universals 80
How do we know which general terms designate universals? • • Roughly: terms used by scientists to designate entities about which we have a plurality of different kinds of testable proposition (cell, electron. . . ) 81
Language has the power to create general terms • which go beyond the domain of universals studied by science 82
Problem: fiat demarcations • male over 30 years of age with family history of diabetes • abnormal curvature of spine • participant in trial #2030 83
Problem: roles • fist • patient • FDA-approved drug • accident 84
Administrative ontologies often need to go beyond universals • Fall on stairs or ladders in water transport injuring occupant of small boat, unpowered Railway accident involving collision with rolling stock and injuring pedal cyclist Nontraffic accident involving motor-driven snow vehicle injuring pedestrian 85
Class =def • • a maximal collection of particulars determined by a general term (‘cell’. ‘electron’ but also: ‘ ‘restaurant in Palo Alto’, ‘Italian’) the class A = the collection of all particulars x for which ‘x is A’ is true 86
universals vs. their extensions universals {a, b, c, . . . } collections of particulars 87
Extension =def • The extension of a universal A is the class: instance of the universal A • (it is the class of A’s instances) • (the class of all entities to which the term ‘A’ applies) 88
Problem • The same general term can be used to refer both to universals and to collections of particulars. Consider: • • HIV is an infectious retrovirus HIV is spreading very rapidly through Asia 89
universals vs. classes universals {c, d, e, . . . } classes 90
universals vs. classes universals defined classes 91
universals vs. classes universals populations, . . . 92
Defined class =def • a class defined by a general term which does not designate a universal • the class of all diabetic patients in Leipzig on 4 June 1952 93
OWL is a good representation of defined classes • sibling of Finnish spy • member of Abba aged > 50 years 94
Terminology =def. • a representational artifact whose representational units are natural language terms (with IDs, synonyms, comments, etc. ) which are intended to designate universals together with defined classes. 95
universals, classes, concepts universals defined classes ‘concepts’ ? 96
universals < defined classes < ‘concepts’ • ‘concepts’ which do not correspond to defined classes: • ‘Surgical or other procedure not carried out because of patient's decision’ • ‘Congenital absent nipple’ • because they do not correspond to anything 97
(Scientific) Ontology =def. • • • a representational artifact whose representational units (which may be drawn from a natural or from some formalized language) are intended to represent 1. universals in reality 2. those relations between these universals which obtain universally (= for all instances) lung is_a anatomical structure lobe of lung part_of lung 98
THE END 99
- Slides: 99