sapiСистемный анализ и прикладная информатика«System analysis and applied information science»2309-49232414-0481Belarusian National Technical University10.21122/2309-4923-2022-4-60-64sapi-595Research ArticleОбработка информации и принятие решенийData processing and decision–makingOptimizing the performance of a server-based classification for a large business document flowOptimizing the performance of a server-based classification for a large business document flowCлавинО. А.SlavinO. A.

Главный научный сотрудник, доктор технических наук

oslavin@isa.ruФедеральный исследовательский центр “Информатика и управление” Российской академии наукРоссияFederal Research Center “Informatics and Management” of the Russian Academy of Sciences; Smart Engines Service LLCRussian Federation202224022023046064Copyright © Cлавин О.А., 20232023Cлавин О.А.Slavin O.A.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://sapi.bntu.by/jour/article/view/595

.

The document categorization problem in the case of a large business document flow is considered. Textual and visual embeddings were employed for classification. Textual embeddings were extracted via OCR Tesseract. The Viola and Jones method was applied to generate visual embeddings. This paper describes the performance optimization technology for the implemented classification algorithm. Servers with Intel CPUs were used for the algorithm execution. For single-threaded implementation, high-level and low-level optimizations were performed. High-level optimization was based on the parametrization of the recognition algorithms and the employment of intermediate data. Low-level optimization was carried out via compiler tools allowing for an extended set of SIMD instructions. The implementation of parallelization with several multithreaded applications on multiple servers was also described. The proposed solution was tested using own test data sets of business documents. The proposed method can be applied in modern information systems to analyze the content of a large flow of digital document images.

text analysisdocument recognitiondocument classificationspeedupReferencesБашкатова, A. Цифровая экономика плодит все больше бумаг: Россияне не скоро перестанут носить в организации справки // Независимая Газета. – 2019 – 14 ноя. [Электронный ресурс] – Режим доступа: https://www.ng.ru/ economics/2019-11-14/4_7727_paper.html, – Загл. с экрана – Яз. рус. Дата доступа – 08.11.2022.Bashkatova, A. Cifrovaya ekonomika plodit vse bol’she bumag: Rossiyane ne skoro perestanut nosit’ v organizacii spravki // Nezavisimaya Gazeta – 2019 – 14 nov. . [Электронный ресурс] – Режим доступа: https://www.ng.ru/economics/2019-11-14/4_7727_ paper.html, – Загл. с экрана – Яз. рус. Дата доступа – 08.11.2022.Liu, L., Wang, Z., Qiu, T., Chen, Q., Lu, Y., Suen, C.Y. Document image classification: Progress over two decades, Neurocomputing 2021, 453: 223-240.Liu, L., Wang, Z., Qiu, T., Chen, Q., Lu, Y., Suen, C.Y. Document image classification: Progress over two decades, Neurocomputing 2021, 453: 223-240.Byun, Y., Lee, Y. Form classification using DP matching. ACM Symposium on Applied Computing 2000; 1: 1–4.Byun, Y., Lee, Y. Form classification using DP matching. ACM Symposium on Applied Computing 2000; 1: 1–4.Devlin, J., Chang, M.-W., Lee, K., Toutanova, K. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. 2019. [Электронный ресурс] – Режим доступа: https://arxiv.org/abs/1810.04805/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.Devlin, J., Chang, M.-W., Lee, K., Toutanova, K. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. 2019. [Электронный ресурс] – Режим доступа: https://arxiv.org/abs/1810.04805/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.Rubin, T.N., Chambers, A., Smyth, P., Steyvers, M. Statistical topic models for multi-label document classification. Machine Learning – 2011, Vol. 88, № 1, 157–208. https://doi.org/10.1007/s10994-011-5272-5.Rubin, T.N., Chambers, A., Smyth, P., Steyvers, M. Statistical topic models for multi-label document classification. Machine Learning – 2011, Vol. 88, № 1, 157–208. https://doi.org/10.1007/s10994-011-5272-5.Vorontsov, K.V., Potapenko, A.A. Tutorial on probabilistic topic modeling: Additive regularization for stochastic matrix factorization. Communications in Computer and Information Science – 2014, Vol. 436, pp. 29-46. https://doi.org/10.1007/978-3-31912580-0_3.Vorontsov, K.V., Potapenko, A.A. Tutorial on probabilistic topic modeling: Additive regularization for stochastic matrix factorization. Communications in Computer and Information Science – 2014, Vol. 436, pp. 29-46. https://doi.org/10.1007/978-3-31912580-0_3.NIST Special Database 2 [Электронный ресурс] – Режим доступа: https://www.nist.gov/srd/nist-special-database-2/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.NIST Special Database 2 [Электронный ресурс] – Режим доступа: https://www.nist.gov/srd/nist-special-database-2/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.Tobacco-3482 [Электронный ресурс] – Режим доступа: https://www.kaggle.com/patrickaudriaz/tobacco3482jpg/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.Tobacco-3482 [Электронный ресурс] – Режим доступа: https://www.kaggle.com/patrickaudriaz/tobacco3482jpg/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.OCR Tesseract [Электронный ресурс] – Режим доступа: https://github.com/tesseract-ocr/tesseract/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.OCR Tesseract [Электронный ресурс] – Режим доступа: https://github.com/tesseract-ocr/tesseract/, – Загл. с экрана – Яз. англ. Дата доступа – 08.11.2022.Tereshin, A.A., Usilin, S.A., Arlazarov, V.V. Performance Improvement of Multi-class Detection Using Greedy Algorithm for Viola-Jones Cascade Selection. Proceedings Volume 10696, Tenth International Conference on Machine Vision (ICMV 2017); 106960D (2018). https://doi.org/10.1117/12.2310101Tereshin, A.A., Usilin, S.A., Arlazarov, V.V. Performance Improvement of Multi-class Detection Using Greedy Algorithm for Viola-Jones Cascade Selection. Proceedings Volume 10696, Tenth International Conference on Machine Vision (ICMV 2017); 106960D (2018). https://doi.org/10.1117/12.2310101Slavin, O.A., Farsobina, V., Myshev, A.V. Analyzing the content of business documents recognized with a large number of errors using modified Levenshtein distance. Cyber-Physical Systems: Intelligent Models and Algorithms. – 2022, Springer Nature Switzerland AG., Vol. 417, pp. 267 – 279. https://doi.org/10.1007/978-3-030-95116-0Slavin, O.A., Farsobina, V., Myshev, A.V. Analyzing the content of business documents recognized with a large number of errors using modified Levenshtein distance. Cyber-Physical Systems: Intelligent Models and Algorithms. – 2022, Springer Nature Switzerland AG., Vol. 417, pp. 267 – 279. https://doi.org/10.1007/978-3-030-95116-0Slavin, O.A. Using Special Text Points in the Recognition of Documents. Studies in Systems, Decision and Control. – 2020, Springer Nature Switzerland AG., Vol 259. pp. 43–53. https://doi.org/10.1007/978-3-030-32579-4_4Slavin, O.A. Using Special Text Points in the Recognition of Documents. Studies in Systems, Decision and Control. – 2020, Springer Nature Switzerland AG., Vol 259. pp. 43–53. https://doi.org/10.1007/978-3-030-32579-4_4Konaka, F., Miura, T. Semantic similarity for sequenced shingles, – 2015 IEEE Pacific Rim Conference on Communications, Computers and Signal Processing (PACRIM), pp. 12-17. https://doi.org/10.1109/PACRIM.2015.7334801.Konaka, F., Miura, T. Semantic similarity for sequenced shingles, – 2015 IEEE Pacific Rim Conference on Communications, Computers and Signal Processing (PACRIM), pp. 12-17. https://doi.org/10.1109/PACRIM.2015.7334801.Acar, U.A., Blelloch, G.E., Harper, R. Selective memorization. ACM SIGPLAN Notices, – 2003, Vol. 38, Issue 1, pp 14–25. https://doi.org/10.1145/640128.604133Acar, U.A., Blelloch, G.E., Harper, R. Selective memorization. ACM SIGPLAN Notices, – 2003, Vol. 38, Issue 1, pp 14–25. https://doi.org/10.1145/640128.604133Tatarowicz, A.L., Curino, C., Jones, E. P. C. and Madden, S. Lookup Tables: Fine-Grained Partitioning for Distributed Databases. – 2012 IEEE 28th International Conference on Data Engineering, pp. 102-113. https://doi.org/10.1109/ICDE.2012.26.Tatarowicz, A.L., Curino, C., Jones, E. P. C. and Madden, S. Lookup Tables: Fine-Grained Partitioning for Distributed Databases. – 2012 IEEE 28th International Conference on Data Engineering, pp. 102-113. https://doi.org/10.1109/ICDE.2012.26

The authors declare that there are no conflicts of interest present.