Vol 44, No 5 (2018)

The relevance of the results of search, classification, and description of the shapes of objects in images largely depends on the quality of vectorization, i.e., on determining regions that are uniform in color and texture and on constructing their boundaries. A color image segmentation algorithm that clusterizes the color palette of the image by constructing the three-dimensional histogram in the color space HSV is proposed. A feature of this algorithm is that it searches the local maximums on the histogram by scanning the color space with a three-dimensional neighborhood analysis operator. Furthermore, an algorithm for approximating boundaries of regions by line segments and circular arcs that recursively augments the approximated chains is proposed. These algorithms are designed for automating the process of extracting informative features from images for the purpose of using them in computer vision systems, content-based image retrieval systems, geographic information systems and other decision support systems based on graphical information.

Currently, there is no doubt among experts in the field of program certification and quality assurance that automated program analysis methods should be used to find bugs that lead to program security vulnerabilities. The national standard for the secure software development requires the use of source code static analysis tools as one of the measures of software quality assurance at the development stage and the application of dynamic analysis and fuzz-testing of the source code at the qualification testing stage. Fundamental limitations of automated program analysis and testing methods make it impossible to carry out simultaneously exhaustive and precise analysis of programs for errors. Thereof, researches are nowadays carried out aimed at reducing the effect of fundamental limitations on the quality and productivity of automated software error detection methods. This paper discusses an approach that combines methods of source code static analysis and dynamic symbolic execution in order to increase the program error detection efficiency.

Contemporary software documentation is as complicated as the software itself. During its lifecycle, the documentation accumulates a lot of “near duplicate” fragments, i.e. chunks of text that were copied from a single source and were later modified in different ways. Such near duplicates decrease documentation quality and thus hamper its further utilization. At the same time, they are hard to detect manually due to their fuzzy nature. In this paper we give a formal definition of near duplicates and present an algorithm for their detection in software documents. This algorithm is based on the exact software clone detection approach: the software clone detection tool Clone Miner was adapted to detect exact duplicates in documents. Then, our algorithm uses these exact duplicates to construct near ones. We evaluate the proposed algorithm using the documentation of 19 open source and commercial projects. Our evaluation is very comprehensive – it covers various documentation types: design and requirement specifications, programming guides and API documentation, user manuals. Overall, the evaluation shows that all kinds of software documentation contain a significant number of both exact and near duplicates. Next, we report on the performed manual analysis of the detected near duplicates for the Linux Kernel Documentation. We present both quantative and qualitative results of this analysis, demonstrate algorithm strengths and weaknesses, and discuss the benefits of duplicate management in software documents.

An approach to user identification based on deviations of their topic trends in operation with text information is presented. An approach is proposed to solve this problem; the approach implies topic analysis of the user’s past trends (behavior) in operation with text content of various (including confidential) categories and forecast of their future behavior. The topic analysis of user’s operation implies determining the principal topics of their text content and calculating their respective weights at the given instants. Deviations in the behavior in the user’s operation with the content from the forecast are used to identify this user. In the framework of this approach, our own original time series forecasting method is proposed based on orthogonal non-negative matrix factorization (ONMF). Note that ONMF has not been used to solve time series forecasting problems before. The experimental research held on the example of real-world corporate emailing formed out of the Enron data set showed the proposed user identification approach to be applicable.