Automatic knowledge extraction from OCR documents using hierarchical document analysis

Masum, Mohammad; Kosaraju, Sai; Bayramoglu, Tanju; Modgil, Girish; Kang, Min-Hee

doi:10.1145/3264746.3264793

Cited by 3 publications

(6 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The maximum values became (359.99, 200.31, 113.71, 359.98, 223.59, 211.74, 189.75, 1399.1). The number of bits to code each bits are (9,8,7,9,8,8,8,11) by taking log2 of the maximum values and round it up. This gives a total of 68 bits for each entry in the dataset.…”

Section: = *mentioning

confidence: 99%

“…Any OCR system is built on finding suitable features of the character images that can be used to classify the image, and then developing the classification algorithm. There are many surveys that explain several feature extraction methods and classification and recognition algorithms [3][4][5][6][7]. The character document image's physical and logical structure, and how its different parts relate, lead to the proper feature extraction method [ 8].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

English Characters OCR Pertinent for Mobile Devices

Al-Shehri¹

2021

IJCDS

View full text Add to dashboard Cite

Optical character recognition concerns translating character images into character text. Four main stages constitute any optical recognition system. They are pre-processing, feature extraction, character recognition, and post-processing. The two most important stages are the feature extraction method and the character recognition algorithm. The optical character recognition system for mobile application must fulfill two requirements, small system size and high speed. In this research, a lightweight footprint feature dataset was created based on the center and centroid of the character image together with other simple image statistics. Fast character recognition algorithm based on weighted Euclidean distance was adapted. A mobile phone application prototype was developed. An accuracy of over 99% was achieved. The execution time for recognizing one character was in millisecond scale when a mobile phone model was used.

show abstract

Section: = *mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

English Characters OCR Pertinent for Mobile Devices

Al-Shehri¹

2021

IJCDS

View full text Add to dashboard Cite

show abstract

“…Different natural language processing models are widely used for knowledge extraction from documents [16]. An efficient framework for knowledge extraction system was proposed that can extract efficient information from thousands of unstructured OCR contract documents [4]. The framework used rule-based methods for converting unstructured OCR documents into hierarchical structed JSON format and then vector space model was utilized for retrieving ranked relevant information from the documents.…”

Section: Related Workmentioning

confidence: 99%

“…Manually relevant knowledge extraction from the large volume of articles is labor-intensive, unscalable and challenging [4,15]. Consequently, there have been several attempts to develop intelligent systems to automatically extract relevant knowledge from many unstructured documents.…”

Section: Introductionmentioning

confidence: 99%

Actionable Knowledge Extraction Framework for COVID-19

Masum

Shahriar

Haddad

et al. 2020

2020 IEEE International Conference on Big Data (Big Data)

View full text Add to dashboard Cite

In response to the COVID-19 pandemic, the White House and a coalition of leading research groups have prepared the COVID-19 Open Research Dataset (CORD-19) containing over 51,000 scholarly articles, including over 40,000 with full text, about COVID-19, SARS-CoV-2, and related coronaviruses. Medical professional including physicians frequently seek answers to specific questions to improve guidelines and decisions. The huge resource of medical literature is important sources to generate new insights that can help medical communities to provide relevant knowledge and overall fight against the infectious disease. There are ongoing attempts to develop intelligent systems to automatically extract relevant knowledge from many unstructured documents. In this paper, we propose an efficient question answering framework based on automatically analyzing thousands of articles to generate both long text answers (sections/ paragraphs) in response to the questions that are posed by medical communities. In the process of developing the framework, we explored natural language processing techniques like query expansion, data preprocessing, and vector space models early. We show the initial results of an example query answering for the incubation period.

show abstract

“…Also handcraft patterns can't understand the complex semantic information in the documents. (M. Mohammad and et al 2018) proposed to construct structural formatted data, by extracting document layout features and then analyzing the changes in the layout features. However, according to practical experience, it is hard to depend on the authors of documents to describe a process strictly following the layout rules.…”

Section: Introductionmentioning

confidence: 99%

A-BPS: Automatic Business Process Discovery Service using Ordered Neurons LSTM

Han

Mei

et al. 2020

2020 IEEE International Conference on Web Services (ICWS)

View full text Add to dashboard Cite

Automatic process discovery from textual process documentations is highly desirable to reduce time and cost of Business Process Management (BPM) implementation in organizations. However, existing automatic process discovery approaches mainly focus on identifying activities out of the documentations. Deriving the structural relationships between activities, which is important in the whole process discovery scope, is still a challenge. In fact, a business process has latent semantic hierarchical structure which defines different levels of detail to reflect the complex business logic. Recent findings in neural machine learning area show that the meaningful linguistic structure can be induced by joint language modeling and structure learning. Inspired by these findings, we propose to retrieve the latent hierarchical structure present in the textual business process documents by building a neural network that leverages a novel recurrent architecture, Ordered Neurons LSTM (ON-LSTM), with process-level language model objective. We tested the proposed approach on data set of Process Description Documents (PDD) from our practical Robotic Process Automation (RPA) projects. Preliminary experiments showed promising results.

show abstract

Automatic knowledge extraction from OCR documents using hierarchical document analysis

Cited by 3 publications

References 25 publications

English Characters OCR Pertinent for Mobile Devices

English Characters OCR Pertinent for Mobile Devices

Actionable Knowledge Extraction Framework for COVID-19

A-BPS: Automatic Business Process Discovery Service using Ordered Neurons LSTM

Contact Info

Product

Resources

About