Machine Learning is the study of methods for programming computers to learn. Computers are applied to a wide range of tasks, and for most of these it is relatively easy for programmers to design and implement the necessary software. However, there are many tasks for which this is difficult or impossible. These can be divided into four general categories. First, there are problems for which there exist no human experts. For example, in modern automated manufacturing facilities, there is a need to predict machine failures before they occur by analyzing sensor readings. Because the machines are new, there are no human experts who can be interviewed by a programmer to provide the knowledge necessary to build a computer system. A machine learning system can study recorded data and subsequent machine failures and learn prediction rules. Second, there are problems where human experts exist, but where they are unable to explain their expertise. This is the case in many perceptual tasks, such as speech recognition, hand-writing recognition, and natural language understanding. Virtually all humans exhibit expert-level abilities on these tasks, but none of them can describe the detailed steps that they follow as they perform them. Fortunately, humans can provide machines with examples of the inputs and correct outputs for these tasks, so machine learning algorithms can learn to map the inputs to the outputs. Third, there are problems where phenomena are changing rapidly. In finance, for example, people would like to predict the future behavior of the stock market, of consumer purchases, or of exchange rates. These behaviors change frequently, so that even if a programmer could construct a good predictive computer program, it would need to be rewritten frequently. A learning program can relieve the programmer of this burden by constantly modifying and tuning a set of learned prediction rules. Fourth, there are applications that need to be customized for each computer user separately. Consider, for example, a program to filter unwanted electronic mail messages. Different users will need different filters. It is unreasonable to expect each user to program his or her own rules, and it is infeasible to provide every user with a software engineer to keep the rules up-to-date. A machine learning system can learn which mail messages the user rejects and maintain the filtering rules automatically. Machine learning addresses many of the same research questions as the fields of statistics, data mining, and psychology, but with differences of emphasis. Statistics focuses on understanding the phenomena that have generated the data, often with the goal of testing different hypotheses about those phenomena. Data mining seeks to find patterns in the data that are understandable by people. Psychological studies of human learning aspire to understand the mechanisms underlying the various learning behaviors exhibited by people (concept learning, skill acquisition, strategy change, etc.).

Machine learning

Document structure analysis, such as zone segmentation and table recognition, is a complex problem in document processing and is an active area of research. The recent success of deep learning in solving various computer vision and machine learning problems has not been reflected in document structure analysis since conventional neural networks are not well suited to the input structure of the problem. In this paper, we propose an architecture based on graph networks as a better alternative to standard neural networks for table recognition. We argue that graph networks are a more natural choice for these problems, and explore two gradient-based graph neural networks. Our proposed architecture combines the benefits of convolutional neural networks for visual feature extraction and graph networks for dealing with the problem structure. We empirically demonstrate that our method outperforms the baseline by a significant margin. In addition, we identify the lack of large scale datasets as a major hindrance for deep learning research for structure analysis and present a new large scale synthetic dataset for the problem of table recognition. Finally, we open-source our implementation of dataset generation and the training framework of our graph networks to promote reproducible research in this direction.

Rethinking Table Recognition using Graph Neural Networks

https://digitalcommons.aaru.edu.jo/cgi/viewcontent.cgi?article=1024&context=fcij

A survey on exploring key performance indicators

Tabular structures in documents offer a complementary dimension to the raw textual data, representing logical or quantitative relationships among pieces of information. In digital mail room applications, where a large amount of administrative documents must be processed with reasonable accuracy, the detection and interpretation of tables is crucial. Table recognition has gained interest in document image analysis, in particular in unconstrained formats (absence of rule lines, unknown information of rows and columns). In this work, we propose a graph-based approach for detecting tables in document images. Instead of using the raw content (recognized text), we make use of the location, context and content type, thus it is purely a structure perception approach, not dependent on the language and the quality of the text reading. Our framework makes use of Graph Neural Networks (GNNs) in order to describe the local repetitive structural information of tables in invoice documents. Our proposed model has been experimentally validated in two invoice datasets and achieved encouraging results. Additionally, due to the scarcity of benchmark datasets for this task, we have contributed to the community a novel dataset derived from the RVL-CDIP invoice data. It will be publicly released to facilitate future research.

/pdf/table-detection-in-invoice-documents-by-graph-neural-4ckddkpud2.pdf

Table Detection in Invoice Documents by Graph Neural Networks

Tables are powerful and popular tools for organizing and manipulating data. A vast number of tables can be found on the Web, which represent a valuable knowledge resource. The objective of this survey is to synthesize and present two decades of research on web tables. In particular, we organize existing literature into six main categories of information access tasks: table extraction, table interpretation, table search, question answering, knowledge base augmentation, and table augmentation. For each of these tasks, we identify and describe seminal approaches, present relevant resources, and point out interdependencies among the different tasks.

Web Table Extraction, Retrieval, and Augmentation: A Survey

Spreadsheet applications are one of the most used tools for content generation and presentation in industry and the Web. In spite of this success, there does not exist a comprehensive approach to automatically extract and reuse the richness of data maintained in this format. The biggest obstacle is the lack of awareness about the structure of the data in spreadsheets, which otherwise could provide the means to automatically understand and extract knowledge from these files. In this paper, we propose a classification approach to discover the layout of tables in spreadsheets. Therefore, we focus on the cell level, considering a wide range of features not covered before by related work. We evaluated the performance of our classifiers on a large dataset covering three different corpora from various domains. Finally, our work includes a novel technique for detecting and repairing incorrectly classified cells in a post-processing step. The experimental results show that our approach delivers very high accuracy bringing us a crucial step closer towards automatic table extraction.

/pdf/a-machine-learning-approach-for-layout-inference-in-2tlhbdup2e.pdf

A Machine Learning Approach for Layout Inference in Spreadsheets

Spreadsheet software are very popular data management tools. Their ease of use and abundant functionalities equip novices and professionals alike with the means to generate, transform, analyze, and visualize data. As a result, spreadsheets are a great resource of factual and structured information. This accentuates the need to automatically understand and extract their contents. In this paper, we present a novel approach for recognizing tables in spreadsheets. Having inferred the layout role of the individual cells, we build layout regions. We encode the spatial interrelations between these regions using a graph representation. Based on this, we propose Remove and Conquer (RAC), an algorithm for table recognition that implements a list of carefully curated rules. An extensive experimental evaluation shows that our approach is viable. We achieve significant accuracy in a dataset of real spreadsheets from various domains.

/pdf/table-recognition-in-spreadsheets-via-a-graph-representation-46a8he0iw5.pdf

Table Recognition in Spreadsheets via a Graph Representation

Spreadsheets are one of the most successful content generation tools, used in almost every enterprise to perform data transformation, visualization, and analysis. The high degree of freedom provided by these tools results in very complex sheets, intermingling the actual data with formatting, formulas, layout artifacts, and textual metadata. To unlock the wealth of data contained in spreadsheets, a human analyst will often have to understand and transform the data manually. To overcome this cumbersome process, we propose a framework that is able to automatically infer the structure and extract the data from these documents in a canonical form. In this paper, we describe our heuristics-based method for discovering tables in spreadsheets, given that each cell is classified as either header, attribute, metadata, data, or derived. Experimental results on a real-world dataset of 439 worksheets (858 tables) show that our approach is feasible and effectively identifies tables within partially structured spreadsheets.

/pdf/table-identification-and-reconstruction-in-spreadsheets-3zpxb19g41.pdf

Table Identification and Reconstruction in Spreadsheets

Spreadsheets compose a notably large and valuable dataset of documents within the enterprise settings and on the Web. Although spreadsheets are intuitive to use and equipped with powerful functionalities, extracting and reusing data from them remains a cumbersome and mostly manual task. Their greatest strength, the large degree of freedom they provide to the user, is at the same time also their greatest weakness, since data can be arbitrarily structured. Therefore, in this paper we propose a supervised learning approach for layout recognition in spreadsheets. We work on the cell level, aiming at predicting their correct layout role, out of five predefined alternatives. For this task we have considered a large number of features not covered before by related work. Moreover, we gather a considerably large dataset of annotated cells, from spreadsheets exhibiting variability in format and content. Our experiments, with five different classification algorithms, show that we can predict cell layout roles with high accuracy. Subsequently, in this paper we focus on revising the classification results, with the aim of repairing misclassifications. We propose a sophisticated approach, composed of three steps, which effectively corrects a reasonable number of inaccurate predictions.

/pdf/cell-classification-for-layout-recognition-in-spreadsheets-2k20hvdkfc.pdf

Cell Classification for Layout Recognition in Spreadsheets

There has been growing interest for more than a decade in Business Analytics as a means for improving business performance. One of the most popular Business Analytics technique involves monitoring performance by means of Key Performance Indicators (KPIs). A KPI is a powerful tool that relates enterprise data to business goals, thereby enabling managers to guide the analytic process and identify deviations in their strategic plan. Nevertheless, monitoring KPIs requires that they are evaluated at multiple levels of detail, in order to identify potential problems earlier instead of being noted after the fact. Unfortunately, there are obstacles to the generation and enactment of such monitoring processes. In particular, there is no systematic, tool-supported process for defining what is to be monitored given a strategic plan, nor are there tools for automatically generating monitoring queries. As a result, monitoring consists of a manual process whereby queries are generated for high level indicators across a few scorecards and dashboards. In this paper we present a systematic semi-automatic approach that covers the entire monitoring process. Our approach performs a partial search guided by the KPIs of the company, generating queries required during the monitoring process. Thanks to our approach, users become aware of the existence of problems and where they are located, without requiring a priori information about the nature of the problem being searched. Moreover, we have implemented our approach in our CASE tool HERMES and evaluated the results on a case study using real data.

/pdf/a-systematic-approach-for-dynamic-targeted-monitoring-of-46fmoejc9i.pdf

Elvis Koci

Papers

A Machine Learning Approach for Layout Inference in Spreadsheets

Table Recognition in Spreadsheets via a Graph Representation

Table Identification and Reconstruction in Spreadsheets

Cell Classification for Layout Recognition in Spreadsheets

A systematic approach for dynamic targeted monitoring of KPIs