A Bank Information Extraction System Based on Named Entity Recognition with CRFs from Noisy Customer Order Texts in Turkish

Emekligil, Erdem; Arslan, Secil; Ağin, Onur

doi:10.1007/978-3-319-45880-9_8

Cited by 9 publications

(5 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Yapılan deneysel çalışmalarda, geliştirilen yazılım, farklı özelliklere sahip 15 adet fiziksel kartvizitteki verileri, %84,76 Doğruluk, %96,05 Kesinlik, %84,88 Duyarlılık, %90,12 F1 Skoru ve ortalama 1,6 sn'lik çıkartım süreleriyle okuyarak ayrıştırabilmektedir. [1], görme engelliler için destek uygulaması [2], bankacılık uygulaması [3], otonom sürüş uygulaması [4] gibi birçok alanda kullanılmaktadır.…”

Section: Otomatik Türkçe Kartvizit Tanıma Için Bulut Tabanlı Web Uygu...unclassified

Cloud Based WEB Application Design for Automatic Turkish Business Card Recognition and Its Performance Evaluation

ŞAHİN

Uçar

Solak

2022

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım Ve Teknoloji

View full text Add to dashboard Cite

In this study, a cloud-based web application is developed that extracts information such as names, phone numbers, e-mail addresses, job titles from physical business cards prepared in Turkish using Tesseract-based optical character recognition method (OCR). Figure A. System architecture of the proposed cloud-based business card reader application Purpose: In the literature, there are various applications developed based on Mobile Vision, OpenCV and Tesseract for the digitization and extraction of information on paper-based business cards in English, Vietnamese, Japanese and Chinese languages. In this study, it is aimed to develop a high-accuracy cloud-based business card recognition software (TRCardScan) compatible with Turkish language with Tesseract-based OCR in order to extract the information from paper-based business cards in Turkish.Theory and Methods: The system architecture of the proposed paper-based Turkish business card reader application is given in Figure A. As can be seen, firstly, business card photos, are taken from the camera or image gallery as input to the application. Then, these photos are subjected to character reading process with Tesseract-based OCR method. As a result of the OCR process, the read and converted texts such as name, surname, mobile phone, e-mail address and contact address on the business card are parsed by algorithms specific to the characteristic of the parts in which it is located. In the last stage, the data that is parsed and transformed into meaningful information is sent to the web service to be written in the relevant field in the database. Results:In the analyzes made with 15 paper-based Turkish business cards with different features, it was observed that the proposed TRCardScan software was able to extract the information from physical business cards with 84.76% accuracy, 96.05% precision, 84.88% recall and 90.12% F1 score. In addition, the average extraction time per business card is 1.6 seconds. Conclusion:The proposed TRCardScan can read and parse data from physical business cards with an average extraction time of 1.6 seconds and high accuracy of around 85%. These results show that the parsing algorithm designed for the Turkish language in the proposed web-based application is successful, considering the time and performance criteria. Finally, when compared to similar software, TRCardScan is considered to be quite successful with its high accuracy rates and reasonable extraction times.

show abstract

Section: Otomatik Türkçe Kartvizit Tanıma Için Bulut Tabanlı Web Uygu...unclassified

Cloud Based WEB Application Design for Automatic Turkish Business Card Recognition and Its Performance Evaluation

ŞAHİN

Uçar

Solak

2022

Gazi Üniversitesi Fen Bilimleri Dergisi Part C: Tasarım Ve Teknoloji

View full text Add to dashboard Cite

show abstract

“…In finance, named-entity recognition (NER) is used for extracting predefined types of data from a document. In banking, transaction order documents of customers may come via fax, which results in very diverse documents because of the lack of a fixed template and creates the need for proper feature extraction to obtain a structured document (Emekligil et al 2016).…”

Section: Information Extractionmentioning

confidence: 99%

Comprehensive review of text-mining applications in finance

et al. 2020

View full text Add to dashboard Cite

Text-mining technologies have substantially affected financial industries. As the data in every sector of finance have grown immensely, text mining has emerged as an important field of research in the domain of finance. Therefore, reviewing the recent literature on text-mining applications in finance can be useful for identifying areas for further research. This paper focuses on the text-mining literature related to financial forecasting, banking, and corporate finance. It also analyses the existing literature on text mining in financial applications and provides a summary of some recent studies. Finally, the paper briefly discusses various text-mining methods being applied in the financial domain, the challenges faced in these applications, and the future scope of text mining in finance.

show abstract

“…The highest accuracy was achieved, ie, 0.82. Emekligil et al used CRF to deal with the defects of different format, complex text, and low resolution of the bank data, and finally realize the entity identification of the customer name, organization name, bank account number, and quantity. Since the concept of deep learning was proposed by Hinton in 2006, it has developed rapidly.…”

Section: Related Workmentioning

confidence: 99%

A co‐training based entity recognition approach for cross‐disease clinical documents

Chen

Che

et al. 2018

Concurrency and Computation

View full text Add to dashboard Cite

Summary Entity recognition plays an important role in building the electronic medical records (EMRs) based medical knowledge graph, which is significant for building Clinical decision support (CDS) system. Cross‐disease clinical documents are context‐related and have different interrelated semantic structures, which bring challenges for entity recognition using traditional methods. In order to solve these problems, this paper proposes a co‐training based entity recognition approach for cross‐disease clinical documents. In this model, we first build partial annotation corpus of the single disease using dependency syntax analysis and the medical statement rule unifies. Then, according to the partial annotation corpus of different diseases, the sentence level features are extracted through the Bi‐LSTM layer with memory unit and CRF methods, which optimize the whole sequence and improve the combination probability of sequence labels. Finally, the results with higher confidence are selected by cross feedback to label the corpus, which enlarges the size of corpus and improves the accuracy of the document entity recognition. The experiment result proves the availability and high efficiency of our method.

show abstract

A Bank Information Extraction System Based on Named Entity Recognition with CRFs from Noisy Customer Order Texts in Turkish

Cited by 9 publications

References 10 publications

Cloud Based WEB Application Design for Automatic Turkish Business Card Recognition and Its Performance Evaluation

Cloud Based WEB Application Design for Automatic Turkish Business Card Recognition and Its Performance Evaluation

Comprehensive review of text-mining applications in finance

A co‐training based entity recognition approach for cross‐disease clinical documents

Contact Info

Product

Resources

About