Automatic problem-solving in archaeology: a computational framework

Joan A. Barceló

«Archeologia e Calcolatori» 1993, 4, 61-80

Abstract

In this paper I have tried to build a computational theory. In other words, a “theory” implemented in a computer program. When using a computational theory we try to solve scientific problems, that is to say, we do not retrieve data units, but we “instantiate” a solution for the problem. I have formalised the concept of an “archaeological problem” in the following way: how is an artefact (or set of artefacts) used by a community in a specific context. The task is then to evaluate the social uses of a specific set of artefacts (Final Situation or State) in terms of: a) their description, and b) all information available about the social, cultural or chronological context and about the human community who produced those artefacts (Initial Situation or State).We may then represent problem solving knowledge as a list of discrete and closed units. Those declarative units are successive states of the problem. We substitute equations for explicit sets of propositions. We can implement a set of answers and a set of decision rules for each one. The resulting program looks like a complex database and not like a mathematical procedure, and we may consider the problem-solving mechanism as a sequential search in a preexisting problem space, using a finite number of particular decision rules. Some interesting work has been done in mathematical representation of archaeological theories, but such approaches have not been very successful, maybe because social sciences cannot be exclusively represented by mathematical models, or because archaeologists are incapable to communicate between themselves using mathematical expressions. As a consequence, archaeologists tend to express their theories by means of linguistic sentences, which is inadequate, given the fact that natural language obstructs objectivity. A representation in terms of logical propositions appears then as the best representation tool available to build social theories. Artificial Intelligence scientists are now exploring this possibility. In this paper I propose an analogy between the structure of archaeological (and social sciences) theories and the mechanism of Turing Machines: given some empirical data (observation of the archaeological record) and a knowledge-base (constituted by high-level concepts and their middle-range correlates), we have to explain the particular case (the archaeological record) by means of the knowledge-base (the theory). The logical mechanism is modus ponens.

Figures

View figures in Interactive Atlas of Digital Images