Semyon Grigorev

Path querying with conjunctive grammars is known to be undecidable. There is an algorithm for path querying with linear conjunctive grammars which provides an over-approximation of the result, but there is no algorithm for arbitrary conjunctive grammars. We propose the first algorithm for path querying with arbitrary conjunctive grammars. The proposed algorithm is matrix-based and allows us to efficiently apply GPGPU computing techniques and other optimizations for matrix operations.

The Bar-Hillel theorem states that context-free languages are closed under intersection with a regular set. This theorem has a constructive proof and thus provides a formal justification of correctness of the algorithms for applications mentioned above. Mechanization of the Bar-Hillel theorem, therefore, is both a fundamental result of formal language theory and a basis for the certified implementation of the algorithms for applications. In this work, we present the mechanized proof of the Bar-Hillel theorem in Coq.

We propose a way to combine formal grammars and artificial neural networks for biological sequences processing. Formal grammars encode the secondary structure of the sequence and neural networks deal with mutations and noise. In contrast to the classical way, when probabilistic grammars are used for secondary structure modeling, we propose to use arbitrary (not probabilistic) grammars which simplifies grammar creation. Instead of modeling the structure of the whole sequence, we create a grammar which only describes features of the secondary structure. Then we use matrix-based parsing to extract features: the fact that some substring can be derived from some nonterminal is a feature. After that, we use a dense neural network to process features.

One of the problems in graph data analysis is querying for specific paths. Such queries are usually performed by means of a formal grammar that describes the allowed edge-labeling of the paths. Path query is said to be calculated using relational query semantics if it is evaluated to triple ((A,v1,v2), such that there is a path from v1 to v2 such that the labels on the edges of this path form a string derivable from the nonterminal A. We focus on the Boolean languages that use Boolean grammars to describe the labeling of paths. Although path querying using relational query semantics and Boolean grammars is known to be undecidable, in this work we propose a path querying algorithm on acyclic graphs which uses relational query semantics and Boolean grammars and approximates the exact solution. To achieve better performance in compare with the naive algorithm, considered classes of graphs were limited to acyclic graphs.

Transparent integration of a domain-specific language for specification of context-free path queries (CFPQs) into a general-purpose programming language as well as static checking of errors in queries may greatly simplify the development of applications using CFPQs. LINQ and ORM can be used for the integration, but they have issues with flexibility: query decomposition and reusing of subqueries are a challenge. Adaptation of parser combinators technique for paths querying may solve these problems. Conventional parser combinators process linear input, and only the Trails library is known to apply this technique for path querying. We demonstrate that it is possible to create general parser combinators for CFPQ which support arbitrary context-free grammars and arbitrary input graphs. We implement a library of such parser combinators and show that it is applicable for realistic tasks.

Extended abstract at TyDe 2018 (at ICFP).

There are several solutions for CFPQ, but how to provide structural representation of query result which is practical for answer processing and debugging is still an open problem. In this paper we propose a graph parsing technique which allows one to build such representation with respect to given grammar in polynomial time and space for arbitrary context-free grammar and graph. Proposed algorithm is based on generalized LL parsing algorithm, while previous solutions are based mostly on CYK or Earley algorithms, which reduces time complexity in some cases.

We present a technique for syntax analysis of a regular set of input strings. This problem is relevant for the analysis of string-embedded languages when a host program generates clauses of embedded language at run time. Our technique is based on a generalization of RNGLR algorithm, which, inherently, allows us to construct a finite representation of parse forest for regularly approximated set of input strings. This representation can be further utilized for semantic analysis and transformations in the context of reengineering, code maintenance, program understanding etc. The approach in question implements relaxed parsing: non-recognized strings in approximation set are ignored with no error detection.