4.4 Discussion
4.4.3 Speech Act Theory
Speech act theories are concerned with providing an informal but precise account of the consequences of an utterance. The study of speech acts was initiated by Austin [8], and has subsequently received much attention (cf. [133]). In this section, we briefly discuss connections between speech act theory and the logic that we have developed. Austin [8] distinguished between three kinds of acts that are associated with an utterance – (a) locutionary acts, (b) illocutionary acts, and (c) perlocutionary acts. The three types of acts are best understood using an example. Suppose John and Mary are getting married in a church, and the priest utters the words “I pronounce you man and wife”. The locutionary act is the construction of sounds in making the utterance. The illocutionary act is that John and Mary are married (according to the church and/or state). And, the perlocutionary act is that “People who witness
the wedding consider John and Mary to be married”. How are these acts reflected in the logic? The locutionary acts are obtained during the creation of utterances via policies (Section 4.3.3). The illocutionary and perlocutionary acts/effects are obtained via inferences, as we discussed in the context of reasoning about rights in Section 4.3.2. Note that for the illocutionary and perlocutionary effects to be obtained in the example above, we need corresponding utterances from the church or state, and the people witnessing the wedding, to the effect that they permit the priest to say (on their behalf) that John and Mary are married. The logic does not distinguish between illocutionary and perlocutionary effects.
Searle [135] classifies illocutionary acts into the following types:
(a) Assertives, which are used to convey the truth of the expressed proposition. For example, “John has good credit”.
(b) Directives, which are used to guide behavior. Examples include commands, advice, etc.
(c) Commissives, which commit the speaker to some future action, e.g., promising or making a bet.
(d) Expressives, which convey an attitude to a proposition. For example, “I am glad that John and Mary are married”.
(e) Declaratives, which change reality in accordance with the proposition of the declarion, e.g., baptisms, pronouncements, etc.
Four of these illocutionary effects that can be achieved using statements in the
logic. Facts are asserted. For example, a branch of a company may assert the cred-
itworthiness of customers. Obligations give rise todirectives. The granting of power,
via permission to speak, is acommissive, as the principal granting the power is com-
mitted to making a statement if the power is excercised. For example, if the state permits a priest to say that John and Mary are married, and the priest says so, then we conclude that the state says that John and Mary are married. And, finally, the
to be married on behalf of the state.
We have connected the logic to speech act theory at a coarse-grained level. An investigation of more fine-grained connections is left to future work.
4.5
Conclusions
We have motivated and described a logic for access control and conformance. The
focus was on the interaction betweensayingandpermission, as needed for these appli-
cations. We proposed two axioms to characterize their interaction (Section 4.3.2), and showed how these axioms could be incorporated into a logic programming approach (Section 4.3.3).
A combined analysis of saying and permission yielded benefits to both applica- tions. For access control, we find a way to avoid the problematic interaction between hand-off and classical reasoning. Our axioms yield a decidable logic with a com- plete semantics (Section 4.3.2), and we hope that they have intuitive appeal to the reader. For conformance, we provide a novel account of recursive notions of power. We showed, in Section 4.3.5, that conformance checking is decidable. In Section 4.3.6, we identified a fragment of the logic, called chain formulas, in which provability can be decided in polynomial time.
We believe that the joint study of access control and conformance is a rich area for research. In Chapter 6, we will identify several avenues for further inquiry.
Chapter 5
Annotating and Computing Logical
Form
5.1
Introduction
Regulatory bodies are large and complex. Manually translating the regulation to logic, as needed for conformance checking, is time-consuming. A long term goal of our work is to use natural language processing (NLP) techniques to assist in this translation. The problem of translating natural language sentences to logic has been of interest for several years in linguistics (cf. [67]), and more recently in NLP [24, 149, 150]. There are many open problems in the design of an appropriate logic and in the translation procedure. In previous chapters, we have motivated and designed features in a logic to accomodate a sentential translation of regulation. In this chapter, we turn our attention to the translation procedure with two objectives:
1. Identifying additional features that are needed in the logic – There are many features that are needed in a logic, to accommodate a sentential translation of regulation. While we have identified some features in previous chapters, an empirical approach is needed to determine whether there are important/frequent constructions which cannot be translated directly.
2. Defining a component of the translation that can be evaluated– Since the logic is a subject of research, it is difficult to make progress on the translation procedure at the same time. We will assume that the end goal is to translate sentences into a variant of the logic that we have developed in previous chapters. The short-term goal is to compute some representation from the text that could serve as an intermediate step in the translation to logic.