Semantic resources (WordNet, Wikidata, BabelNet, ... ) offer invaluable knowledge that can be exploited by humans and machines to solve a variety of tasks. Among these, we address here the one called entity set expans...
详细信息
ISBN:
(纸本)9783319733050;9783319733043
Semantic resources (WordNet, Wikidata, BabelNet, ... ) offer invaluable knowledge that can be exploited by humans and machines to solve a variety of tasks. Among these, we address here the one called entity set expansion: extend a given a set of words-called seeds- with new ones being of the same "sort". Differently from classical approaches, we determine "optimal" common categories of the given seeds by analyzing the semantic relations among the objects these seeds refer to. In particular, we define the notion of an entity network to integrate information from different semantic resources, and show how to use such networks to disambiguate word senses. Finally, we propose a proof-of-concept implementation in answer set programming with external predicates to query online semantic resources and perform optimization tasks.
Several AI problems can be conveniently modelled in ASP, and many of them require to enumerate solutions characterized by an optimality property that can be expressed in terms of subset-minimality with respect to some...
详细信息
ISBN:
(纸本)9783031157073;9783031157066
Several AI problems can be conveniently modelled in ASP, and many of them require to enumerate solutions characterized by an optimality property that can be expressed in terms of subset-minimality with respect to some objective atoms. In this context, solutions are often either (i) answersets or (ii) sets of atoms that enforce the absence of answersets on the ASP program at hand such sets are referred to as minimal unsatisfiable subsets (MUSes). In both cases, the required enumeration task is currently not supported by state-of-the-art ASP solvers.
Implementation costs linked to processor memory subsystems (cache miss costs, stalls due to bandwidth limits, etc.) have been shown to be a factor in the performance of a variety of declarative programming tools. This...
详细信息
Implementation costs linked to processor memory subsystems (cache miss costs, stalls due to bandwidth limits, etc.) have been shown to be a factor in the performance of a variety of declarative programming tools. This article investigates their impact on answerset solvers and the factors that control them. Experiments independently altering the size and difficulty of input programs allow a qualitative assessment of whether input program or solver design is a greater factor and a quantitative assessment of how much of problem these issues create.A variety of processor performance metrics are recorded and used to provide a detailed picture of what limits solver performance and dispel a number of common *** demonstrate the degree to which these problems can be addressed, smodels-ie is presented. This is a version of the smodels solver with a number of implementation changes to improve cache utilisation, one major aspect of memory costs.
Reasoning in the presence of imprecision and vagueness is inevitable in many real-world applications including those in robotics and intelligent agents. Although, reasoning about actions is a major component in these ...
详细信息
ISBN:
(纸本)9789899507968
Reasoning in the presence of imprecision and vagueness is inevitable in many real-world applications including those in robotics and intelligent agents. Although, reasoning about actions is a major component in these real-world applications, current actions languages for reasoning about actions lack the ability to represent and reason about actions in the presence of imprecision and vagueness that stem from effects of actions in these real-world applications. In this paper we present a new action language called fuzzy action language, A(F), that allows the representation and reasoning about actions with vague (fuzzy) effects. In addition we define the notions of fuzzy planning and fuzzy plan in the fuzzy action language AF. Furthermore, we describe a fuzzy planner based on the fuzzy action language AF that is developed by translating a fuzzy action theory, FT, in A(F) into a normal logic program with answerset semantics, II, where trajectories in FT are equivalent to the answersets of II. In addition, we formally prove the correctness of our translation. Furthermore, we show that fuzzy planning problems can be encoded as a SAT problem.
The design of new products is often an evolutionary process, where product versions are built on one another. This form of (PGE) reuses some parts of previously developed systems, while others have to be designed from...
详细信息
ISBN:
(纸本)9783031045806;9783031045790
The design of new products is often an evolutionary process, where product versions are built on one another. This form of (PGE) reuses some parts of previously developed systems, while others have to be designed from scratch. In consideration of subsequent design steps, i.e., verification, testing, and production, PGE may significantly reduce the time-to-market as these steps can be skipped for reused parts. Thus, deciding which components have to be replaced or added to meet the updated requirements while preserving as many legacy components as possible is one of the key problems in PGE. A further aspect of PGE is the potentially more efficient search for valid design candidates. An already optimized base system can be systematically extended by new functionality without the necessity to search the entire design space. To this end, in this work, we propose a systematic approach, based on answer set programming, to exploit the ideas of PGE in electronic system-level design space exploration. The idea is to gather information on a previous design, analyze the changes to a new version, and utilize the information to steer the search towards potentially good regions in the design space. Extensive experiments show that the presented approach is capable of finding near-optimal design points up to 1,000 times faster than a conventional approach.
With the rapid growth in research on and development of autonomous machines, machine ethics, which used to be "just a theory", has gained greater practical importance. In this paper, we present a logical app...
详细信息
ISBN:
(纸本)9783030649838;9783030649845
With the rapid growth in research on and development of autonomous machines, machine ethics, which used to be "just a theory", has gained greater practical importance. In this paper, we present a logical approach to machine ethics. Our objective is to enable autonomous machines to behave in morally appropriate ways following well-defined ethical principles, exercising sound ethical judgement. Since moral reasoning involves selecting appropriate behavioural actions with varying preconditions, we propose a non-monotonic reasoning model and encode the model through two types of well-known ethical frameworks: the consequentialist approach to ethics and the deontological approach to ethics. The computational model is developed using answer set programming in a situation calculus framework. We apply our model to a few paradigmatic scenarios that can be encountered in autonomous driving and interactions with social robots. Our study shows that the proposed model is applicable to a wide range of scenarios and leads to appropriately different reasoning outputs in different ethical frameworks.
Processable English (PENG) is a controlled natural language designed to specify and conceptualize knowledge in a human-readable and machine-processable way. PENG specifications can be translated unambiguously into exe...
详细信息
ISBN:
(纸本)9781450366038
Processable English (PENG) is a controlled natural language designed to specify and conceptualize knowledge in a human-readable and machine-processable way. PENG specifications can be translated unambiguously into executable answerset programs (ASP). In this paper we suggest an extension of the language PENG ASP so that weak constraints can be expressed in controlled natural language and processed as part of an ASP program. In contrast to strong constraints that have always to be satisfied and are already part of the controlled natural language, we introduce weak constraints that can be weighted and prioritised and should be satisfied whenever possible. The addition of weak constraints to the controlled natural language PENG ASP makes it possible to specify optimisation problems in a natural way.
We present plingo, an extension of the ASP system clingo with various probabilistic reasoning modes. Plingo is centered upon LPMLN, a probabilistic extension of ASP based on a weight scheme from Markov Logic. This cho...
详细信息
ISBN:
(纸本)9783031215407;9783031215414
We present plingo, an extension of the ASP system clingo with various probabilistic reasoning modes. Plingo is centered upon LPMLN, a probabilistic extension of ASP based on a weight scheme from Markov Logic. This choice is motivated by the fact that the core probabilistic reasoning modes can be mapped onto optimization problems and that LPMLN may serve as a middle-ground formalism connecting to other probabilistic approaches. As a result, plingo offers three alternative frontends, for LPMLN, P-log, and ProbLog. The corresponding input languages and reasoning modes are implemented by means of clingo's multi-shot and theory solving capabilities. The core of plingo amounts to a re-implementation of LPMLN in terms of modern ASP technology, extended by an approximation technique based on a new method for answerset enumeration in the order of optimality. We evaluate plingo's performance empirically by comparing it to other probabilistic systems.
In this paper, we present a syntactic transformation, called the unfolding operator, that allows forgetting an atom in a logic program (under ASP semantics). The main advantage of unfolding is that, unlike other synta...
详细信息
ISBN:
(纸本)9783031157073;9783031157066
In this paper, we present a syntactic transformation, called the unfolding operator, that allows forgetting an atom in a logic program (under ASP semantics). The main advantage of unfolding is that, unlike other syntactic operators, it is always applicable and guarantees strong persistence, that is, the result preserves the same stable models with respect to any context where the forgotten atom does not occur. The price for its completeness is that the result is an expression that may contain the fork operator. Yet, we illustrate how, in some cases, the application of fork properties may allow us to reduce the fork to a logic program, even in conditions that could not be treated before using the syntactic methods in the literature.
We present FOLD-SE, an efficient, explainable machine learning algorithm for classification tasks given tabular data containing numerical and categorical values. The (explainable) model generated by FOLD-SE is represe...
详细信息
ISBN:
(纸本)9783031520372;9783031520389
We present FOLD-SE, an efficient, explainable machine learning algorithm for classification tasks given tabular data containing numerical and categorical values. The (explainable) model generated by FOLD-SE is represented as a set of default rules. FOLD-SE uses a novel heuristic called Magic Gini Impurity for literal selection that we have devised. FOLD-SE uses a refined data comparison operator and eliminates the long tail effect. Thanks to these innovations, explainability provided by FOLD-SE is scalable, meaning that regardless of the size of the dataset, the number of learned rules and learned literals stay quite small while good accuracy in classification is maintained. Additionally, the rule-set constituting the model that FOLD-SE generates does not change significantly if the training data is slightly varied. FOLD-SE is competitive with state-of-the-art traditional machine learning algorithms such as XGBoost and Multi-Layer Perceptrons (MLP) w.r.t. accuracy of prediction while being an order of magnitude faster. However, unlike XGBoost and MLP, FOLD-SE generates explainable models. The FOLD-SE algorithm outperforms prior rule-learning algorithms such as RIPPER in efficiency, performance, and scalability, especially for large datasets. FOLD-SE generates a far smaller number of rules than earlier algorithms that learn default rules.
暂无评论