Abstract:

Baikal Natural Territory (BNT) is the territory that adjacent to Lake Baikal, which is a unique natural object and, in accordance with the UNESCO Convention, a "World Natural Heritage". Baikal is in the central part of the Baikal Rift Zone (BRZ) – the most active seismic zone located in the middle of Russia. The development of the BRZ leads to the emergence of dangerous geological processes that can lead to a violation of the balance in the Lake Baikal ecological system and the surrounding area. In addition, these processes and phenomena pose a real threat to the smooth functioning of mainline communications, hydroelectric power plants and strategically important industries in the region, which, according to the classification of the Ministry of Emergency Situations of Russia, belongs to the first category of danger. To ensure systematic monitoring and forecasting of the environmental situation of the BNT, systematic observations are organized, as well as obtaining and analysing information about the activity of hazardous geological processes in digital form. The digital transformation of monitoring of hazardous geological processes, resulting from the digitalization of processes and the development of appropriate infrastructure, provides the possibility of using new models and methods, more flexible approaches to the analysis of ongoing processes and the prediction of possible extreme events. In this paper, a digital platform is proposed that provides support for the digital transformation of the monitoring of hazardous geological processes using the example of BNT. The platform under consideration may be used for ecological monitoring of BNT area.

Abstract:

This article addresses the problem of quantifying the confidence of large language models (LLMs) in the automatic semantic classification of arguments with emotive predicates. Using Russian-language social media posts, we analyze verbs of fear (to scare, to be afraid, etc.) and emotional attitude (to like, to love) with the semantic roles of experiencer, causator, and object. The study aims to compare self-assessed confidence of LLM Claude Sonnet 4.5 with expert assessments of the model's reasoning texts in argument classification in the healthcare domain. The experiment utilized a stratified set of 300 examples using chain-of-thoughts reasoning in Russian and four-step confidence scale. The results showed a moderate Spearman correlation between the expert and model assessments. A statistically significant relationship was found only between self-assessment of the model and the actual classification accuracy, while expert assessment of the linguistic characteristics of reasoning was unrelated to accuracy. It was concluded that explicit LLM reasoning is not directly related to its self-assessment of confidence and is separate from the decision-making process; reasoning may be an important functional part of the user interface, but not of the research.

Abstract:

A method for assessing the risk of default of a corporate borrower at the monitoring stage based on a scoring assessment is proposed. This paper provides proof of the hypothesis that scoring methods for assessing credit risks can be used not only at the stage of initial assessment of a potential borrower when making a decision on lending, but also at the stage of its monitoring when accompanying a transaction. Monitoring is a periodic review of the credit quality of the corporate borrower with whom the loan agreement is concluded. This is done for the purpose of timely detection of negative signals, as well as timely response to threatening trends in the borrower's activities.

Some credit institutions save on monitoring by relying on the decision-making system, considering it flawless. However, this saving can be a fatal mistake, since many things change over time during the "life" of the enterprise. This is facilitated by both external factors (political, economic) and internal (incorrect development strategy of the organization, inability to assess its own credit capabilities, unscrupulous counterparties).

The proposed method is a system of automatic risk signals that have been tested for predictive ability, excluding manual procedures. The proposed solution includes markers (risk signals) that have a predictive ability above average, which can lead to a default of the corporate borrower. In addition, color marking is applied – red, yellow, green, which allows you to visualize the criticality of the identified risk signal depending on the predictive ability - a visual representation of the borrower's risks in order to facilitate interpretation.

The analysis of the developed method showed how much it is possible to speed up the monitoring process, which will allow for a prompt response to the identified risk signals, as well as to predict the likely deterioration of the borrower's credit quality in the loan or guarantee portfolio without compromising the quality of risk assessment.

Abstract:

The paper addresses the formal representation of knowledge about disinformation in a low-resource bilingual media environment. Automatic fake news detection is traditionally formulated as a binary REAL/FAKE classification task, which does not capture the internal structure of an unreliable message: the type of fake content, the disinformation technique employed, the author’s communicative intent, and the characteristics of the source and evidence base.

Two interrelated resources are presented. The first is KazFakeCorpus, a bilingual corpus of 4,276 texts in Kazakh and Russian, balanced by class within each language; the REAL class was drawn from official materials of the Gov.kz portal and the FAKE class from their controlled transformations. The texts were annotated in Label Studio by two independent experts following a multi-level semantic scheme; Krippendorff’s Alpha for the principal levels ranged from 0.79 to 0.88. The second resource is KazFakeOnto, an ontology that recasts the annotation scheme from a flat relational representation into a formal OWL 2 DL model: 38 classes, 26 object properties, 22 data properties, 112 individuals and 1,155 asserted triples. Eleven defined classes are not asserted manually but inferred by a reasoner from property values; a property chain links a news item to its annotator through an annotation object. Consistency was verified with HermiT, and the query layer was evaluated on fifteen SPARQL and thirteen DL queries.

The ontological layer is shown to provide a reproducible specification of the annotation scheme, the automatic inference of semantic categories, and a unified formal environment for formulating corpus-linguistic questions and analysing machine-learning model errors. The limitations of the approach and directions for scaling the ontology to the full corpus and to naturally occurring disinformation are discussed.

Abstract:

This paper examines the challenges of assessing the reliability of information extracted from popular science texts. Popular science is known to be characterized by a free style of presentation, loose references to primary sources and expert opinions, and the presence of controversial statements and conclusions. Clarifying these components of the author's reasoning can significantly increase the credibility among critical readers or, conversely, refute the authors' assertions.

This paper proposes a new approach to assessing the reliability of information contained in popular science publications. This approach utilizes D. Walton's argumentation theory to identify potentially unreliable claims and logical fallacies. Searching scientific databases and calculating trust metrics for scientific publications based on their metadata and scientific ratings will enable the verification of controversial claims and an assessment of their reliability. The use of large language models will enable the generation of explanations for decisions made.

To support the development and validation of this approach, two data collection methods are proposed. The first uses existing corpora of Russian-language popular science texts and commentaries with annotated argumentation. The annotation includes, among other things, polemical argumentation schemes that challenge the author's thesis or argument. The second method utilizes LLM for generating synthetic theses and popular science articles, based on actual scientific publications.

Abstract:

A method for building end-to-end management accounting in a division of the Bank’s subdevision specializing in working with non-core assets is proposed. Has been proposed the process approach, an algorithm for building a database for the formation of key performance and control indicators.

Has been described the key stages of the department's work, the attribute composition of entities (set) arriving, enriched and transmitted at each stage of the department's work. By modeling the process has been built a role model, access and editing rights for employees. Data sources (reference books) for optimization and unification of the process of filling the database (tuple) are proposed. A method of accessing the database in the Power Query Microsoft Excel add-in is proposed, which allows you to collect data from files of all basic data types, process and refine the received data. In the interactive programming environment Jupyter Notebook, mathematical and financial models for data analysis (logistic regression, decision tree and discounted cash flow method) were built based on data in order to predict costs, the timing of asset exposure and make a decision on the optimal cost of putting property on the Bank's balance sheet and selling price. Based on ready-made libraries (matpotlib, seaborn, plotly), options for data visualization for management are proposed. Using the example of the Bank's division, the author describes the positive effects and opportunities that open up to the management of different levels in solving day-to-day tasks and planning the activities of the division. A technical task was proposed for the development of a showcase for the sale of non-core assets on the Bank's website as an environment for the accumulation of external data for making flexible management decisions.

Abstract:

The project approach is widely used in organizing the training of IT specialists in universities. Despite the fact that the organized development processes are extremely close to the processes used in commercial IT companies, the analysis of the development process of student projects is not carried out. This article proposes an approach for performing analytics of the development process in a student project, and also discusses options for using the results of analytics in assessing student work.

Abstract: Research on design and development of software and technological support for geo-portal of ICM SB RAS are discussed. Its main components and implementation details are presented. A number of problems are discussed in details, such as web-based metadata catalog, the logic of building applications based on web services geoportal. A list of implemented information systems, based on discussed technologies is presented. The authors were directly involved in the develop-ment and implementation of geoportal based projects. In developing the software many different software libraries and components were used. Web mapping user interface was created using a number of open source libraries. To create a server-side web application authors used GIS platforms MapGuide Open Source and Min-nesota Mapserver. GeoWebCache was another essential component of distributed web mapping ap-plications. By analyzing and summarizing the experience gained creating information systems, it should be noted that the use of geoportal-based solutions in this area, can dramatically improve the efficiency of software development and problems solving.

Abstract: International Virtual Observatory (IVO) is a collection of integrated astronomical data archives and software tools that utilize computer networks to create an environment in which research can be conducted. Several countries have initiated national virtual observatory programs that will combine existing databases from ground-based and orbiting observatories and make them easily accessible to researchers. As a result, data from all the world's major observatories will be available to all users and to the public. This is significant not only because of the immense volume of astronomical data but also because the data on stars and galaxies have been compiled from observations in a variety of wavelengths: optical, radio, infrared, gamma ray, X-ray and more. Each wavelength can provide different information about a celestial event or object, but also requires a special expertise to interpret. In a virtual observatory environment, all of this data is integrated so that it can be synthesized and used in a given study. The International Virtual Observatory Alliance (IVOA) represents 17 international projects working in coordination to realize the essential technologies and interoperability standards necessary to create a new research infrastructure. Russian Virtual Observatory is one of the founders and important members of the IVOA. The International Virtual Observatory project was launched about ten years ago, and major IVO achievements in science and technology in recent years are discussed in this presentation. Standards for accessing large astronomical data sets were developed. Such data sets can accommodate the full range of wavelengths and observational techniques for all types of astronomical data: catalogues, images, spectra and time series. The described standards include standards for metadata, data formats, query language, etc. Services for the federation of massive, distributed data sets, regardless of the wavelength, resolution and type of data were developed. Effective mechanisms for publishing huge data sets and data products, as well as data analysis toolkits and services are provided. The services include source extraction, parameter measurements and classification from data bases, data mining from image, spectra and catalogue domains, multivariate statistical tools and multidimensional visualization techniques. Development of prototype VO services and capabilities implemented within the existing data centers, surveys and observatories are also discussed. We show that the VO has evolved beyond the demonstration level to become a real research tool. Scientific results based on end-to-end use of VO tools are discussed in the presentation.

Abstract:

The article presents an information and analytical system for assessing eight cognitive functions of aviation personnel based on fuzzy logic. The system receives 14 input variables, which are the results of subjects passing 13 validated psychodiagnostic methods. The system operates based on the Mamdani fuzzy inference algorithm, scalable to the input space dimension (1, 2 or 3 metrics) and using adaptive bases of production rules (3, 9 or 27). The generated quantitative assessments of eight cognitive functions in the range [0; 1] are fed to a three-level classifier, which implements their qualitative interpretation by assigning them to one of three zones: risk [0;0.4], acceptable level (0.4;0.7] or high level (0.7;1]. The practical significance of the developed information-analytical system lies in its role as a tool for automated decision support for professional selection specialists. Implementing the system will enable a shift from a binary «fit/unfit» assessment to a differentiated approach to managing personnel risks that takes into account the structure of a candidate's cognitive profile.

Abstract: The use of pointers and indirect memory accesses in the program, as well as the complex control flow are some of the main weaknesses of the static analysis of programs. The program properties investigated by this analysis are too conservative to accurately describe program behavior and hence they prevent parallel execution of the program. The application of dynamic analysis allows us to expand the capabilities of semi-automatic parallelization. In the SAPFOR system (System FOR Automated Parallelization), a dynamic analysis tool has been implemented, based on on the instrumentation of the LLVM representation of an analyzed program, which allows the system to explore programs in both C and Fortran programming languages. The capabilities of the static analysis implemented in SAPFOR are used to reduce the overhead program execution, while maintaining the completeness of the analysis. The use of static analysis allows to reduce the number of analyzed memory accesses and to ignore scalar variables, which can be explored in a static way. The developed tool was tested on performance tests from the NAS Parallel Benchmarks package for C and Fortran languages. The implementation of dynamic analysis, in addition to traditional types of data dependencies (flow, anit, output), allows us to determine privitizable variables and a possibility of pipeline execution of loops. Together with the capabilities of DVM and OpenMP these greatly facilitates program parallelization and simplify insertion of the appropriate compiler directives.

Abstract:

This paper presents a digital platform for complex monitoring data of dangerous geodynamic, engineering-geological and hydrogeological processes occurring in the region of intensive nature management of the central ecological zone of the Baikal natural territory (CEZ BNT). The platform is intended for integration and analysis of data coming from several polygons located within the CEZ BPT in order to assess the state of the geological environment and forecasting of hazardous processes manifestation.

The platform is built on a client-server architecture. Storage, processing and analysis of data is carried out on the server, which users can access via the Internet using a web browser. Several data filtering methods (linear frequency, Savitsky-Goley and others), various methods of spectral and wavelet analysis, multifractal and entropy analysis, and spatial data analysis are currently available. The digital platform has been tested on real data.

Abstract: This study discusses some of the shortcomings of Gaussian humanization and discusses how the articulation models developed by drummers can be emulated using a probabilistic model. An author's algorithm for “revitalizing” midi-parts of percussion instruments was developed. Various dependencies and regularities, which are manifested when playing drums, are programmed. The dependence of the notes parameters not only on the parameters of the previous notes, but also on the subsequent notes of the corresponding parts of the drum set was created. The dependence of the impact force on the position of the note in the measure is created. Thus, emphasizing notes in a strong lobe. Also, the dependence of the note's volume on the coincidence with the notes of other parts of the drum set was created. To increase the dynamism and liveliness of the party, the volume of the hat was increased before the impact on the small drum. A comparison was made of the amplitudes of the corresponding notes of the parts, as far as the party, animated using a certain method, is different from the one played by a professional drummer. When listening to tests, batches processed using a modified method objectively sound lively than using Gaussian and quantized batches.

Abstract:

The paper is devoted to the study of the cognitive function associated with the generation of elliptical sentences in the Russian language. The study is conducted by testing this cognitive ability using a computer system specially developed by the authors for this purpose. Testing of this cognitive ability is proposed and implemented for the first time. The system is an extension of Moodle and is openly hosted in the github repository. Elliptical constructions are limited to verbal and nominal ellipses, which are theoretically possible to be completely reconstructed based on the context of the sentence. The study is conducted with the participation of SPbPU students as respondents. The texts of planimetric tasks are chosen as the subject area. As a result of the analysis of testing data, the following results are obtained: the influence of the respondent’s knowledge of the subject area (planimetry) on the test results is established; a tendency towards self-study of respondents was discovered, which is manifested in a reduction in time and an increase in scores as they pass tests; it is shown that respondents are poorly motivated if they do not see feedback on the answer to the completed task. The paper discusses the problems of further development of the testing system and its use in adapting questionnaires (tasks) to assess the knowledge of SPbPU students in the field of automation of bug detection in programs, as well as for diagnosing the functional state of operator specialists and express diagnosis of dementia. It also seems promising to use the system to improve the processes of syntactic parsing of elliptic sentences and automate the restoration of ellipses in the subject area of planimetry.

Abstract:

We investigated the current models of the publication and dissemination of scientific knowledge. We describe the modern information management system of scientific publications and services that determine their functionality.

We discuss the concept of the digital infrastructure of the electronic scientific journal. Under this infrastructure, we understand the complex that combines management software platform of electronic journal and a number of specialized information systems. The software platform realizes the basic operating log management processes. Information systems provide the operation of additional services, taking into account the specifics of the journal subject area.

We present an approach to the organization of the digital infrastructure of the scientific journal based on an open platform Open Journal Systems (OJS). We provide software services that extend the functionality of this platform and considering specificity of the subject area of scientific journals. We have created software modules for automating of electronic scientific journal workflow. These modules are an extension of OJS.

We present a system of services for the automated processing of collections of scientific documents. These services provide verification of document compliance to the accepted rules of formation of collections and their conversion to the established formats; structural analysis of documents and extraction of metadata, as well as their integration into the scientific information space. The system allows to automatically performing a set of operations that cannot be realized for acceptable time with the traditional manual processing of electronic content. It is designed for the large collections of scientific documents.

Algorithms style validation of texts at the article registration stage in the information system of electronic scientific journal, the selection of reviewers, alert and control the timing of reviewing were automated. Information gathering algorithm with dedicated news lines of scientific journals, further analysis and distribution of news by categories and degrees of importance were developed. The algorithm automatically extract bibliographic data from a homogeneous array of publications (in particular, the issues of the scientific journal) and the formation of metadata blocks for export to international information and analytical system were created. Methods integration of OJS platform and international databases of science citation were developed.

We present methods of processing documents containing mathematical formulas: in the collections of documents that contain mathematical formulas, algorithm for the search formulas is developed; basic ideas, approaches and results already obtained by the mathematical knowledge management based on ontology are presented; a method of constructing recommender systems based on mathematical knowledge ontologies described. The method of primary processing automated of scientific article using TеX-notation developed.

The new direction of researches of scientific communications in the environment of Web 2.0 – altmetrics – is considered. We have analyzed the content of the notion «altmetrics», we conducted a comparison of traditional (bibliometric and scientometric) and alternative indicators. We describe the use of world experience informetric services on scientific journals sites. We discussed options for implementing these approaches to create an electronic scientific journal management platform. 

Abstract: Catalogues are traditionally used as a basis for Earth observation (EO) data infrastructures, and enable informational support for large amount of data archived. Multiple Access Centre for Regional Satellite Monitoring of Environment FEB RAS focuses on using techniques, technologies and tools developed under European initiatives of creating global EO infrastructures to provide an access to informational and processing services for satellite data. A number of last year projects developed or promoted by European Space Agency (ESA) use service-oriented platform of SSE (Services Support Environment). This papers discusses some means used in these projects to create EO data catalogues, and integrate them into united distributed heterogeneous network. Experience of using this means in the Center is presented.

Abstract:

The paper addresses the problem of constructing interpretable vector representations of scientific texts for intellectual academic genealogy. A typology of embeddings is proposed, comprising three classes: statistical, learned neural, and structured symbolic. The study argues for combining the strengths of neural embeddings (high semantic accuracy) with those of symbolic embeddings (interpretable dimensions). To operationalize this hybrid approach, an algorithm for learned symbolic embeddings is introduced, which utilizes a regression-based mapping from a model’s internal representation to an interpretable vector of scores.

The approach is evaluated on a corpus of fragments from dissertation abstracts in pedagogy. A compact transformer encoder with a regression head was trained to reproduce topic relevance scores produced by a state-of-the-art generative language model. A comparison of six training setups (three regression-head architectures and two encoder settings) shows that fine-tuning the upper encoder layers is the primary driver of quality improvements. The best configuration achieves R² = 0.57 and a Top-3 accuracy of 74% in identifying the most relevant concepts. These results suggest that, for tasks requiring formalized output representations, a compact encoder with a regression head can approximate a generative model’s behavior at substantially lower computational cost. More broadly, the further development of algorithms for constructing learned symbolic embeddings contributes to building a model of formal knowledge representation in which the convergence of neural and symbolic methods ensures both the scalability of scientific text processing and the interpretability of vector representations that encode their content.

Abstract:

In this work we present the first morphologically annotated corpus for Iron Ossetic that conforms to the Universal Dependencies schema. The corpus includes 5454 manually annotated sentences from the Iron Ossetic Corpus of Oral Texts, containing 74032 tokens. We use this corpus to train a BERT-based morphological analyzer. The analyzer achieves tag accuracy of 95.60%. Furthermore, the paper presents the results of experiments aimed at improving classification quality. It is shown that neither filtering the model’s output using a context‑free analyzer nor multi‑task approach lead to a significant improvement. Finally, the paper presents the results of testing the analyzer on out‑of‑domain data and shows that the model achieves tag accuracy of 91.45% on them.

Abstract:

The technology of modelling various situations in virtual environment systems, which are computer three-dimensional models of a real or artificial environment, is discussed. The user can view these scenes directly on the computer screen, wall screen, in a stereo glasses, virtual reality glasses, etc. He can also move inside a virtual scene and interact with its objects. In turn, the environment can also change. This allows modelling of various situations (situation modelling) in the virtual environment system. With such modelling, some static or dynamic situation is set in the virtual environment system in which the operator must perform the tasks assigned to him.

A mechanism for setting situations by changing a virtual three-dimensional scene using configuration files and virtual control panels is proposed. A special language has been developed for writing configuration files, and a special editor has been developed for creating virtual control panels. The approbation of the proposed methods is presented on the examples of two virtual scenes: a training ground for mobile robots and a jet backpack for the rescue of an astronaut in outer space.

Abstract:

This paper is devoted to the design and development of a system for automating numerical similarity assessment of Android applications. The task of application similarity evaluation is reduced to the similarity evaluation of sets of control flow graphs constructed based on code from classes.dex files of applications. The similarity value was calculated based on the similarity matrix. The algorithms of graph editing and Levenshtein distance were used to compare control flow graphs. Application similarity criteria were formulated and their representation forms were investigated. Types of Android application models and methods of their construction are presented. A prototype of the system for automating the numerical evaluation of Android-applications similarity is developed. Optimization of the software solution is performed with the help of parallel programming tools. Experiments are carried out and the conclusion is made about the ability of the developed system to detect similarities between Android applications.

Abstract:

On example of recent multi-risk disasters and their impacts on electricity transmission grids this paper shows the need of multi-risk assessment and governance. It discusses factors of vulnerability of electricity transmission infrastructure in Europe in lights of changing for architecture of electricity grids requirements, growing volumes of renewable energy sources and probabilities for cascading effects. The paper also brings examples of multi-risk assessment and shows algorithm of calculation for multi-risk disasters.

Abstract:

Amidst the rapid growth of e-commerce and increasing demand for personalization, the Russian market for customized clothing faces a shortage of technological and widely accessible solutions. This paper presents the results of a research and implementation project focused on developing a multi-brand digital platform with an integrated 3D configurator, aimed at transforming the pre-order cycle. The solution enables customers to interactively create garment designs in a web environment, while allowing designers to optimize logistics and minimize overproduction.

The primary scientific and technical contribution of this work lies in its detailed description of the platform's target architecture and a scalable 3D model processing pipeline that ensures model optimization and correct browser-based rendering. An additional contribution is the developed methodology for preparing and optimizing 3D garment models for web visualization. Formalized as a set of technical requirements, this methodology achieves a balance between visual quality and performance.

As a result of this research, the authors have addressed the challenge of unifying 3D model formats from different designers within a multi-brand digital platform—a key distinction from existing single-brand solutions. Furthermore, the implemented technology enables the customization of 3D clothing models with interactive real-time visualization of all design modifications on a single screen.

The technological feasibility and effectiveness of the solution are substantiated by a comparative analysis of existing alternatives, a market analysis using the PAM-TAM-SAM-SOM model, and an assessment of functional requirements.

The article also outlines a practical strategy for implementing the digital platform, making it a valuable resource for researchers and practitioners working at the intersection of e-commerce, computer graphics, and the digital transformation of business processes.

Abstract:

This paper examines modern approaches to subword tokenization of texts as applied to the low-resource Tajik language, which is characterized by a complex morphological structure and a high degree of word-form variability. In the course of the study, a large-scale heterogeneous corpus was compiled and preprocessed, comprising 99 books and 134,497 textual articles of various genres and topics, with a total volume exceeding 33 million tokens. The corpus was cleaned of noise, normalized, and used as a basis for training and subsequent testing of subword models.

Based on this corpus, five tokenization models implementing the BPE, WordPiece, and Unigram algorithms were trained and analyzed using the Hugging Face Tokenizers and SentencePiece libraries. Comparative evaluation was conducted using a set of key metrics, including the proportion of out-of-vocabulary (OOV) words, the degree of text representation compression, tokenization speed, as well as characteristics of n-gram distribution, which make it possible to assess the ability of the models to capture the morphological and structural organization of the language. The experimental results made it possible to identify the strengths and weaknesses of different approaches to subword segmentation and to determine the most effective tokenization strategies under conditions of the morphological complexity of the Tajik language. The findings obtained can be used in the development of language models and applied NLP tools for Tajik and other low-resource languages, contributing to the expansion of their presence in the digital environment.

Abstract:

The use of automatic text processing methods, including full-text description classification methods, allows achieving a significant reduction in labor costs when processing experimental data. This paper discusses the use of the automatic text classification method in the field of processing and classifying core elements and determining lithofacies. Lithofacies are coeval geological bodies (deposits) that differ in composition or structure from adjacent layers. When assessing the oil and gas potential of fields, it is necessary to construct maps and diagrams of lithofacies distribution. This requires classifying a large number of full-text descriptions of core sections prepared by specialists. The algorithm presented in the article allows, based on specified rules and dictionaries, to conduct classification taking into account the order and significance of keywords in sentences. The advantages of this approach are: the ability to distinguish between close lithofacies, the ability to use archival data, ease of adjustment to new classes, adaptation to Russian-language core descriptions and the possibility of local use without the need to transfer core descriptions to third-party applications.

Abstract:

The paper addresses the problem of multi-level similarity analysis of Android applications based on static features in digital application collections. Such collections may contain duplicates, forks, repackaged builds, and other modified variants; malicious payloads are treated as a special case of modification rather than as a synonym of repackaging. The paper formulates a similarity function for Android applications, introduces a static application model as the working object of comparison, and presents a multi-level pipeline that separates candidate screening, in-depth pairwise analysis, result interpretation, and a decision layer. Meaningful similarity signals are sought not only in classes.dex bytecode, but also in AndroidManifest.xml, resources, APK-internal metadata, and library dependencies. A numerical similarity score is computed only when static models are built successfully; otherwise the pipeline records a dedicated technical failure status together with a normalized failure reason. Preliminary evidence is reported on a local pilot set of five core pairs and two boundary cases. These results indicate that explicit handling of shared library code may improve interpretability, but they do not yet constitute a full validation of the proposed architecture on large collections.