A Semiotic Approach to Conceptual Modelling Antonio L. Furtado
Departamento de Informática Pontifícia Universidade Católica do Rio de Janeiro
People make history Charles W. Bachman “the programmer as navigator” IDS, CODASYL, pointers, machines, files
Edgar F. Codd “the casual user” logic, algebra, tables as abstract data types
Peter P. Chen “modeling the things in the real world” diagrams, pictograms, concepts
Our project at PUC-Rio
from databases to information systems logic programming formalism prototype tools Theory, Linked Data: Marco A. Casanova Big Data: Karin K. Breitman Human-Computer Interaction: Simone D. J. Barbosa Digital Entertainment: Bruno Feijó, Angelo E. M. Ciarlini, Cesar Pozzer
Colleagues
Marco Casanova
Bruno Feijó
Simone Barbosa
Angelo Ciarlini
Karin Breitman
Cesar Pozzer
Main aspects • • • •
Three-schemata specifications The plan-recognition / plan-generation paradigm Application domains and narrative genres Semiotic completeness
Three-schemata specifications
static schema facts – entities existing in a given state of the world and their properties (attributes, relationships) dynamic schema events – state transformations caused by the execution of operations defined on their pre-/post-conditions (the STRIPS method) behavioural schema agents – who execute operations in order to reach states wherein their goals are satisfied (agent profiles, goal-inference rules)
Basis: the Entity-Relationship model for facts and everything else – pre- / post-conditions, and goals are expressed in terms of facts
Plan-recognition / plan-generation paradigm
decision-making at each state: how agents evaluate current risks or opportunities – drives (choice of goals), attitudes (choice of plans), emotions (commitment or not)
multi-agent environments: positive and negative goal and plan interferences, competition, cooperation and negotiation (cf. R. Willensky)
plan generation algorithm: pre-conditions of operation O1 fulfilled by post-conditions of operation O2 – recursive backward-chaining process to generate sequences of operations to achieve the agents’ goals
simulation: through the use of the planner, the three-schemata specifications become executable
the reuse alternative: plan recognition algorithm to match observed actions against a library of plan patterns – recognition and adaptation as a form of case-based reasoning
Application domains and narrative genres
Business information systems – application domains with a predefined repertoire of operations (e.g. banking) – The IDB prototype – 1. Prolog, 2. Prolog + Oracle via ODBC, 3. running Oracle environment (operations compiled into stored procedures)
Digital entertainment – narrative genres limited by conventions and a finite number of typical agents and events (Vladimir Propp – characterization of folktales via 7 dramatis personae and 31 functions) – The Logtell prototypes – operational specification of genres – interactively generated plans ⇒ narrative plots dramatization of the generated plots: via natural language, animation, videobased composition, etc.
the reuse alternative: plan-recognition working on types and motifs (cf. Aarne-Thompson’s Index)
Revisiting the ANSI/SPARC proposal •
•
conceptual modelling also at the external (users’) level two reminders from Semiotics: - communication: Jakobson’s message encoding/decoding - representation: Peirce’s interpretant
•
from semantic to pragmatic: cooperative query-response interfaces with access to user profiles and plan-recognition
The Web era: all of us as navigators
Tim Berners-Lee
Recall x Precision
or: The Dangers of Navigation
Recall x Precision
or: The Dangers of Navigation
Our quest for “completeness”
Opportunities and risks at the Web era: recall x precision Focusing on one starting point (e.g. an entity instance), in what direction(s) should we usefully go to find more about it? Charting the information space to obtain some navigation guidelines Completeness – are we missing something?
A word of caution: completeness is relative – world models are by definition simplifications – ours give an operational definition, aiming to mirror some real world application domain (or narrative genre)
Computational completeness - power equivalent to a Turing machine LISP – basic functions
Böhm-Jacopini theorem –
constructor: cons projections: car, cdr predicates: eq, atom
3 control structures are enough: • sequence • selection • iteration
LISP – data structure list – of atoms and/or lists
control in LISP
to represent both sequences and sets and also: graphs, hypergraphs ...
function composition cond recursion
Relational completeness - enough to translate first-order logic expressions relational algebra operations product join projection union difference intersection selection division
Relational completeness - enough to translate first-order logic expressions relational algebra operations product join projection union difference intersection selection division
the five basic operations product projection union selection difference join, intersection, division can be defined in terms of the basic operations
Relational completeness - intuitively: enough for handling tables colums = sequence of domains (horizontal axis) . product . projection rows = set of tuples (vertical axis) . union . selection what else? – removal of a given set of tuples (topological bounds) (serving to achieve the effects of intersection and division, and to perform deletion)
. difference _________________________________________________________ Extension: N2F2 tables:
cells = structured values (depth axis) . nest . unnest
Peter Chen:
logic level (values with no semantics) ⇒ conceptual level
columns
⇒ entity or relationship class attributes
horizontal axis
rows
⇒ instances of the same class
vertical axis
structured values ⇒ semantic hierarchies
depth axis
removal of tuples ⇒ consistent world state maintenance
topological bounds
(integrity constraints, business rules, conventions)
The information space – semiotic relations connectivity vertical axis similarity depth axis hierarchy topological bounds negation horizontal axis
– syntagmatic relations – paradigmatic relations – meronymic relations – antithetic relations
Saussure - Cours de Linguistique Générale (posthumous publication in 1916) Winston, Chaffin and Herrmann – A Taxonomy of Part-Whole Relations Horn – A Natural History of Negation
Connectivity: RDF diagrams "…RDF can be viewed as a member of the Entity-Relationship model family" [Chen, P.P. "Entity-Relationship Modeling: Historical Events, Future Trends, and Lessons Learned". Software pioneers. Springer, 2002].
On the relevance of the other relations as navigation guidelines
similarity – comparison with alternatives, classification criteria, typical instances (Lakoff), similarity measures, clustering, inter-domain analogy, adaptation and reuse hierarchy – zooming in and out, detailed description versus summarization, recognition (gestalt perception), modular design negation – disambiguation, what is not wanted, counter examples, near miss (Winston),opposition, contrary values: Boolean or in a multi-graded scale, conflict, blend (Turner), different opinions, inconsistency, integrity violation, “things that should not be but nevertheless are”
⇒ plentiful recall, with more precision: “if” and “only if”
An aside: theorem-proving methods by inference
connectivity
syntagmatic
by analogy
similarity
paradigmatic
by case analysis
hierarchy
meronymic
by contradiction
negation
antithetic
Semiotic completeness - the four master tropes relation syntagmatic paradigmatic meronymic antithetic
meaning connectivity similarity hierarchy negation
operator and or part-of not
trope metonymy metaphor synecdoche irony
Quintilian, Ramus, Vico, Kenneth Burke, Hayden White, ... Jonathan Culler: “... a system, indeed the system, by which the mind comes to grasp the world conceptually in language” Jean François Champollion – Grammaire égyptienne, 1836
Synecdoche
Metonymy
Metaphor
Enigma
Events and the four relations (plots: post-conditions → pre-conditions) relation syntagmatic paradigmatic meronymic antithetic
meaning coherence alternatives detailed action disruption
trope metonymy metaphor synecdoche irony
Syntagmatic relations
abduct
rescue
elope
capture
Paradigmatic relations
abduct
rescue
elope
capture
Antithetic relations
abduct
rescue
elope
capture
Meronymic relations
abduct
capture
ride
defeat
seize
carry
Meronymic relations
rescue
elope
ride
defeat
entreat
carry
Agents and the four relations (goal & plan positive and negative interferences) relation syntagmatic paradigmatic meronymic antithetic
meaning collaborating, helper parallel, peer collective, fellowship competing, villain
trope metonymy metaphor synecdoche irony
Next step: semiotic diachronic completeness Facts + (events, agents, plans, plots) ⇒ temporal dimension
Provide a narrative viewpoint – plots as entity property Revisit the past – log with time-stamped records of events Add near future forecast – scheduled events also in the log Watch for evil attempts – recognition from observed actions Find what is achievable, and how to get there – plan trials to reach possible (and supposedly impossible / illegal!!!) goals
Concluding remarks
Conceptual modelling in the large and in the small: - from the closed-world of databases to the open-world of the Web - the mini-world of each of us, under a flurry of portable devices
Goals and plans: towards pragmatic user-centered conceptual modelling Contribution of Semiotics Contribution of Human-Computer Interaction (HCI)
