Search

Equation discovery, also known as symbolic regression, is a machine learning task of inducing closed-form equations from data and background knowledge. The latter takes various forms. Domain-specific knowledge can constrain the space of candidate equations to those that make sense in the scientific or engineering domain of use. Cross-domain knowledge, on the other hand, imposes general rules for model acceptability, such as parsimony, understandability, or consistency of the equations with the dimensional units of the variables. In this paper, we propose using attribute grammars to ensure the induced equations' dimensional consistency. Attribute grammars are flexible enough to combine cross-domain knowledge on dimensional consistency with domain-specific knowledge expressed as a probabilistic context-free grammar. At the same time, we show that attribute grammars can be efficiently transformed into probabilistic context-free grammars for equation discovery with existing algorithms. Finally, we provide empirical evidence that attribute grammars ensuring dimensional consistency of equations can significantly improve the performance of equation discovery on the standard set of a hundred Feynman benchmarks. • Attribute grammars elegantly express dimensionally consistent expressions. • Efficient combining of dimensional analysis with other background knowledge. • Generate equations by transforming into probabilistic context-free grammar. • Discovered 60% more benchmark equations by accounting for dimensional consistency. [ABSTRACT FROM AUTHOR]

Product reviews are of great commercial value for online shopping market. The identification of customer opinions from product reviews is helpful to improve the marketing decisions of customers, sellers and producers. This paper proposes a novel framework for summarizing customer opinions from product reviews. Firstly, our framework identifies grammatically and semantically meaningful phrases which contain product attributes and their corresponding opinions from original product reviews by using grammar rules and the latent Dirichlet allocation (LDA) model. Secondly, our framework generates readable and simple summaries from the identified phrases automatically by using the deep neural network. The summaries provide users the valuable opinions on product attributes. Moreover, our framework provides an interactive mode for users to choose product attributes which they are interested for generate personalized summaries to help users focus on the most concerned opinions. Experimental results on six datasets demonstrate effectiveness of our framework. [ABSTRACT FROM AUTHOR]

In this paper we introduce a novel, context-free grammar, RNAFeatures * , capable of generating any RNA structure including pseudoknot structures (pk-structure). We represent pk-structures as orientable fatgraphs, which naturally leads to a filtration by their topological genus. Within this framework, RNA secondary structures correspond to pk-structures of genus zero. RNAFeatures * acts on formal, arc-labeled RNA secondary structures, called λ -structures. λ -structures correspond one-to-one to pk-structures together with some additional information. This information consists of the specific rearrangement of the backbone, by which a pk-structure can be made cross-free. RNAFeatures * is an extension of the grammar for secondary structures and employs an enhancement by labelings of the symbols as well as the production rules. We discuss how to use RNAFeatures * to obtain a stochastic context-free grammar for pk-structures, using data of RNA sequences and structures. The induced grammar facilitates fast Boltzmann sampling and statistical analysis. As a first application, we present an O ( n log ( n )) runtime algorithm which samples pk-structures based on ninety tRNA sequences and structures from the Nucleic Acid Database (NDB). Availability: the source code for simulation results is available at http://staff.vbi.vt.edu/fenixh/TPstructure.zip . The code is written in C and compiled by Xcode. [ABSTRACT FROM AUTHOR]

Comparing XML documents with XML grammars, also known as XML document and grammar validation, is useful in various applications such as: XML document classification, document transformation, grammar evolution, XML retrieval, and the selective dissemination of information. While exact (Boolean) XML validation has been extensively investigated in the literature, the more general problem of approximate (similarity-based) XML validation, i.e., document–grammar similarity evaluation, has not yet received strong attention. In this paper, we propose an original method for measuring the structural similarity between an XML document and an XML grammar (DTD or XSD), considering their most common operators that designate constraints on the existence, repeatability and alternativeness of XML elements/attributes (e.g., ?, ∗ , MinOccurs , MaxOccurs , etc.). Our approach exploits the concept of tree edit distance, introducing a novel edit distance recurrence and dedicated algorithms to effectively compare XML documents and grammar structures, modeled as ordered labeled trees. Our method also inherently performs exact validation by imposing a maximum similarity threshold (minimum edit distance) on the returned results. We implemented a prototype and conducted several experiments on large sets of real and synthetic XML documents and grammars. Results underline our approach’s effectiveness in classifying similar documents with respect to predefined grammars, accurately detecting document and/or grammar modifications, and performing document and grammar relevance ranking. Time and space analysis were also conducted. [ABSTRACT FROM AUTHOR]