Alexandre Miguel Pinto

Revised stable models – a semantics for logic programs

This paper introduces an original 2-valued semantics for Normal Logic Programs (NLP), which conservatively extends the Stable Model semantics (SM) to all normal programs. The distinction consists in the revision of one feature of SM, namely its treatment of odd loops, and of infinitely long support chains, over default negation. This single revised aspect, addressed by means of a Reductio ad Absurdum approach, affords a number of fruitful consequences, namely regarding existence, relevance and top-down querying, cumulativity, and implementation. The paper motivates and defines the Revised Stable Models semantics (rSM), justifying and exemplifying it. Properties of rSM are given and contrasted with those of SM. Furthermore, these results apply to SM whenever odd loops and infinitely long chains over negation are absent, thereby establishing significant, not previously known, properties of SM. Conclusions, further work, terminate the paper.

Layered Models Top-Down Querying of Normal Logic Programs

For practical applications, the use of top-down query-driven proof-procedures is essential for an efficient use and computation of answers using Logic Programs as knowledge bases. Additionally, abductive reasoning on demand is intrinsically a top-down search method. A query-solving engine is thus highly desirable. The current standard 2-valued semantics for Normal Logic Programs (NLPs), the Stable Models (SMs) semantics, does not allow for top-down query-solving because it does not enjoy the relevance property -- and moreover, it does not guarantee the existence of a model for every NLP. To overcome these current limitations we introduce here a new 2-valued semantics for NLPs--the Layered Models semantics -- which conservatively extends the SMs, enjoys relevance and guarantees model existence among other useful properties. Moreover, for existential query answering there is no need to compute total models, but just the partial models that sustain the answer to the query, or one might simply know a model one exists without producing it; relevance ensures these can be extended to total models. A first implementation of a query-solving engine based on this new semantics is presented and described here. It uses the XSB-Prolog engine and its XASP interface to Smodels, thereby providing a useful tool built as a hybrid of the two systems and taking advantage of the best of each. Conclusions and further work end the paper.

Approved models for normal logic programs

We introduce an original 2-valued semantics for Normal Logic Programs (NLPs) extending the well-known Argumentation work of Phan Minh Dung on Admissible Arguments and Preferred Extensions. In the 2-valued Approved Models Semantics set forth, an Approved Model (AM) correspond to the minimal positive strict consistent 2-valued completion of a Dung Preferred Extension. The AMs Semantics enjoys several non-trivial useful properties such as (1) Existence of a 2-valued Model for every NLP; (2) Relevancy, and (3) Cumulativity. Crucially, we show that the AMs Semantics is a conservative extension to the Stable Models (SMs) Semantics in the sense that every SM of a NLP is also an AM, thus providing every NLP with a model: a property not enjoyed by SMs. Integrity constraints, written in a simpler way, are introduced to identify undesired semantic scenarios, whilst permitting these to be produced nevertheless. We end the paper with some conclusions and mention of future work.

Inspecting side-effects of abduction in logic programs

In the context of abduction in Logic Programs, when finding an abductive solution for a query, one may want to check too whether some other literals become true (or false) as a consequence, strictly within the abductive solution found, that is without performing additional abductions, and without having to produce a complete model to do so. That is, such consequence literals may consume, but not produce, the abduced literals of the solution.We show how this type of reasoning requires a new mechanism, not provided by others already available. To achieve it, we present the concept of Inspection Point in Abductive Logic Programs, and show, by means of examples, how one can employ it to investigate side-effects of interest (the inspection points) in order to help choose among abductive solutions. We show how to implement inspection points on top of already existing abduction solving systems -- ABDUAL and XSB-XASP -- in a way that can be adopted by other systems too.

Incremental Answer Completion in the SLG-WAM

The SLG-WAM of XSB Prolog soundly implements the Well-Founded Semantics (WFS) for logic programs, but in a few pathological cases its engine treats atoms as undefined that are true or false in WFS. The reason for this is that the XSB does not implement the SLG Answer Completion operation in its engine, the SLG-WAM --- rather Answer Completion must be performed by post-processing the table. This engine-level omission has not proven significant for applications so far, but the need for Answer Completion is becoming important as XSB is more often used to produce well-founded residues of highly non-stratified programs. However, due to its complexity, care must be taken when adding Answer Completion to an engine. In the worst case, the cost of each Answer Completion operation is proportional to the size of a program P , so that the operation must be invoked as rarely as possible, and when invoked the operation must traverse as small a fragment as possible of P . We examine the complexity of Answer Completion ; and then describe its implementation and performance in XSBs SLG-WAM such that the invocations of the operation are restricted, and which is limited in scope to Strongly Connected Components within a tabled evaluations Subgoal Dependency Graph.

Comparing the Performance of Different NLP Toolkits in Formal and Social Media Text

Nowadays, there are many toolkits available for performing common natural language processing tasks, which enable the development of more powerful applications without having to start from scratch. In fact, for English, there is no need to develop tools such as tokenizers, part-of-speech (POS) taggers, chunkers or named entity recognizers (NER). The current challenge is to select which one to use, out of the range of available tools. This choice may depend on several aspects, including the kind and source of text, where the level, formal or informal, may influence the performance of such tools. In this paper, we assess a range of natural language processing toolkits with their default configuration, while performing a set of standard tasks (e.g. tokenization, POS tagging, chunking and NER), in popular datasets that cover newspaper and social network text. The obtained results are analyzed and, while we could not decide on a single toolkit, this exercise was very helpful to narrow our choice.

Tight Semantics for Logic Programs

We define the Tight Semantics (TS), a new semantics for all NLPs complying with the requirements of: 2-valued semantics; preserving the models of SM; guarantee of model existence, even in face of Odd Loops Over Negation (OLONs) or infinite chains; relevance cumulativity; and compliance with the Well-Founded Model. When complete models are unnecessary, and top-down querying (a la Prolog) is desired, TS provides the 2-valued option that guarantees model existence, as a result of its relevance property. Top-down querying with abduction by need is rendered available too by TS. The user need not pay the price of computing whole models, nor that of generating all possible abductions, only to filter irrelevant ones subsequently. A TS model of a NLP P is any minimal model M of P that further satisfies P^---the program remainder of P---in that each loop in P^ has a MM contained in M, whilst respecting the constraints imposed by the MMs of the other loops so-constrained too. The applications afforded by TS are all those of Stable Models, which it generalizes, plus those permitting to solve OLONs for model existence, plus those employing OLONs for productively obtaining problem solutions, not just filtering them (like Integrity Constraints).

Stable model implementation of layer supported models by program transformation

For practical applications, the use of top-down query-driven proof-procedures is convenient for an efficient use and computation of answers using Logic Programs as knowledge bases. A 2-valued semantics for Normal Logic Programs (NLPs) allowing for top-down query-solving is thus highly desirable, but the Stable Models semantics (SM) does not allow it, for lack of the relevance property. To overcome this limitation we introduced in [11], and summarize here, a new 2-valued semantics for NLPs -- the Layer Supported Models semantics -- which conservatively extends the SM semantics, enjoys relevance and cumulativity, guarantees model existence, and respects the Well-Founded Model. In this paper we exhibit a space and time linearly complex transformation, TR, from one propositional NLP into another, whose Layer Supported Models are precisely the Stable Models of the transform, which can then be computed by extant Stable Model implementations, providing a tool for the immediate generalized use of the new semantics and its applications. TR can be used to answer queries but is also of theoretical interest, for it may be used to prove properties of programs. Moreover, TR can be employed in combination with the top-down query procedure of XSB-Prolog, and be applied just to the residual program corresponding to a query (in compliance with the relevance property of Layer Supported Models). The XSB-XASP interface then allows the program transform to be sent to Smodels for 2-valued evaluation.

Side-Effect Inspection for Decision Making

In order to decide on the course of action to take, one may need to check for side-effects of the possible available preferred actions. In the context of abduction in Logic Programs, abducible literals may represent actions and assumptions in the declarative rules used to represent our knowledge about the world. Besides finding out which alternative sets of actions achieve the desired goals, it may be of interest to identify which of those abductive solutions would also render true side-effect literals relevant for the decision making process at hand, and which would render those side-effects false. After collecting all the alternative abductive solutions for achieving the goals it is possible to identify which particular actions influence inspected side-effect literals’ truth-value.

Describing the Internet of Things with an ontology: The SusCity project case study

The Internet of Things comprises a network of physical objects, like sensors and actuators, collecting and exchanging data. Given the importance of the information exchanged in these environments, the communication infrastructure becomes a critical point that needs to be managed optimally, while providing high-performance levels to end users. To guarantee the correct interaction between the different procedures intended for the optimization of the communication infrastructure, a standard and flexible representation of the data related to the network is necessary. Many of the services could be monitored through the web, thus using standard Web languages with a rich expressive power, such as the languages used in the Semantic Web, would allow for the reification of interoperable descriptions. This paper presents an ontology for the Internet of Things infrastructure tailored to the needs of Smart Cities. Furthermore, different kinds of evaluations were performed to corroborate the correctness of this ontology, including potential infrastructure optimization objectives like low latency and high resilience.