JAROSLAV MÜLLER: TL – JAZYK PRO MULTIAGENTNÍ SYSTÉMY

Transparentní intenzionální logika (TIL) by mohla najít dobré uplatnění v oblastech umělé inteligence a multiagentních systémů. Brání tomu zejména to, že TIL je `filozofická‘ logika – konstrukce se nedají zapsat jako text, chybí něteré praktické záležitosti…

Proto definujeme jazyk TL (TIL Language). TL je počítačový jazyk TILu, definuje formální syntax zapisovatelnou v textovém (ascii) formátu. Rozšiřuje TIL o některé praktické prvky jako jsou sekvence, makra a podpora ontologií.

Zamýšlené využití jazyka TL je jako sémantický jazyk pro komunikaci agentů v multiagentních systémech. Proto TL navazuje na specifikace FIPA, kde by mohl zaujmout roli komunikačního prostředku mezi strojem a člověkem. Zároveň pracujeme na implementaci TL a jeho integraci do prostředí Jade, které implementuje standardy FIPA.

BJÖRN JESPERSEN: MALFUNCTION AND MODIFICATION

Björn Jespersen
TU Delft, Faculty of Technology, Policy and Management, Section of Philosophy
Leiden University, Faculty of Philosophy

Is a broken corkscrew a corkscrew? In general, if x is a malfunctioning F-device, is x then an F-device? The proper-function theorist says Yes, for the proper function of x as an F overrides its malfunctioning as an F. The pragmatist says No, for nothing that fails to function as an F is an F, while anything that does function as an F is an F. I favour an answer in the affirmative, basically because x was designed to function as an F-device. This is not to say, though, that everything that is designed to function as an F is an F. A plan for designing F-devices may be structurally flawed in such a way that nothing manufactured in accordance with it could possibly function as an F-device and, therefore, would not be an F. Our intuitions as to what makes x an F in the first place will strongly influence our intuitions concerning malfunction.

The purpose of my talk is to outline a logical system within which to reason about malfunction. I am going to address the following three issues.

  • formation of the property denoted by the predicate ‘is a malfunctioning F’
  • predication of ‘is a malfunctioning F’ of individual technical artifacts
  • validity, or invalidity, of various arguments in which the predicate ‘is a malfunctioning F’ occurs.

By arguing that a broken corkscrew is still a corkscrew, I have argued that the predicate modifier ‘malfunctioning’ is subsective [roughly, FG(x) / G(x)], hence not privative [roughly, FG(x) / not G(x)]. However, since ‘malfunctioning’, as it occurs in ‘is a malfunctioning F’, is a modifier, it cannot be intersective [roughly, FG(x) / F(x) et G(x)]. Still, we can infer that a broken corkscrew is broken, thanks to the rule of pseudo-detachment, so that we obtain something equivalent to the first conjunct of F(x) et G(x). The rule says, roughly: FG(x) / *F(x). In words, “a malfunctioning F malfunctions”. The idea behind the rule is to convert an attributive occurrence of ‘malfunctioning’, e.g., ‘F’, into a predicative occurrence, ‘F*’. The rule of pseudo-detachment has been developed together with Pavel Materna and Marie Duží. It forms part of a general project on intensional logic, which is in this case applied to properties and the phenomenon of predicate modification.

VÝJEZDNÍ ZASEDÁNÍ GRANTU V PECI P. SNĚŽKOU 18.-20. LISTOPADU 2005

PROGRAM
Pátek 18. 11.

příjezd

19:00-… Radek Honzík: Forcing – aneb vše, co jste chtěli vědet o teorii mnozin a báli jste se zeptat

Sobota 19. 11.

9:00-11:00 Juraj Hvorecký (FLÚ AV ČR), Jan Kolář: Pojmy a filozofia mysle

11:00-12:00 organizační záležitosti

odpolední výlet

16:00-18:00 Ondřej Tomala, Laco Koreň: Logika a sémantika singulárních termínů

Neděle 20. 11.

9:00-11:00 Marta Bílková: Interpolace v modálnich logikách

ROSTISLAV HORČÍK: STANDARDNÍ ÚPLNOST LOGIKY PIMTL

Seminar bude spise technickeho charakteru a bude pojednavat o reseni jednoho otevreneho problemu z vyrokove fuzzy logiky.

Vety o uplnosti ve fuzzy logice vetsinou rikaji, ze dana logika je uplna vzhledem k nejake tride algeber, ktera obvykle tvori varietu. Nicmene ve fuzzy logice casto bereme jako typickou (standardni) mnozinu pravdivostnich hodnot realny interval [0,1]. Z tohoto duvodu se u jednotlivych fuzzy logik zkouma jeste tzv. standardni uplnost, t.j. uplnost vuci tride algeber jejichz nosicem je prave interval [0,1]. Z pohledu algebry je standardni uplnost ekvivalentni s tvrzenim, ze varieta algeber, vuci kterym je logika uplna, je generovana podtridou algeber s nosicem [0,1]. A prave o dukazu vety o standardni uplnosti logiky PiMTL bude prednaska pojednavat.

LIBOR BĚHOUNEK: VÝROKOVÁ FUZZY LOGIKA OTÁZEK

(This is the syllabus of my talk at 8th VlaPoLo, Zielona Gora, on a similar topic.)

The many-valued approach to erotetic logic is one of the most important generalizations of its classical variants. Its value for applications is shown, i.a., by the fact that many questionnaires use scaled answers rather than simple yes-no ones. A specific branch of many-valued logic is fuzzy logic, which is aimed at capturing the notion of vagueness and degrees of truth. Recent advances in metamathematics of fuzzy logic ([1], [2]) made it possible to generalize various branches of mathematics and logic so as to deal with vagueness and uncertainty.

Fuzzy logic takes the set of truth-values (degrees of truth) to be the real interval [0,1]. T-norm based fuzzy logic starts from a few natural assumptions about the truth function representing conjunction: commutativity, associativity, continuity, monotonicity, and classical values for classical arguments. These conditions lead to Hajek’s Basic Fuzzy Logic BL with its important extensions, incl. Goedel and Lukasiewicz infinite-valued logics.

Groenendijk-Stokhof erotetic logic (GS) is an intensional system based on the notion of logical space ([3], [4]). The intension of a declarative sentence is a subset of the logical space; questions are identified with partitions of the logical space. The notions of answerhood and entailment for questions are defined by means of the intensions of the answers and the blocks of the partition.

In the talk, I shall sketch the generalization of GS to fuzzy setting. Starting with fuzzy intensional semantics for propositions (the intension of a proposition is a fuzzy subset of the logical space), I proceed to definitions of fuzzy questions and fuzzy answerhood, entailment and equivalence conditions. I discuss various options for the definitions and show their respective motivation and mutual relationship. Finally, I try to assess the prospects of fuzzified GS in both theory and applications.

References

[1] Hajek, Petr: Metamathematics of Fuzzy Logic. Kluwer, Dordrecht 1998.
[2] Hajek, Petr and Godo, Lluis: Deductive systems of fuzzy logic. Tatra Mountains Math. Publ. 13 (1997), 35-66.
[3] Groenendijk, Jeroen and Stokhof, Martin: Questions. In: van Benthem and ter Meulen (eds.), Handbook of Logic and Language, Elsevier/MIT 1994.
[4] Groenendijk, Jeroen and Stokhof, Martin: Partitioning Logical Space. 2nd ESSLLI Annotated Handout, Leuven 1990.