Extraction of Address Data from Unstructured Text using Free Knowledge Resources

The majority of data on computers nowadays is in the form of unstructured data and unstructured text. The inherent ambiguity of natural language makes it incredibly difficult but also highly profitable to find hidden information or comprehend complex semantics in unstructured text. In this paper, we present the combination of natural language processing (NLP) and convolution neural network (CNN) hybrid architecture called automated analysis of unstructured text using machine learning (AAUT-ML) for the detection of complex semantics from unstructured data that enables different users to make understand formal semantic knowledge to be extracted from an unstructured text corpus. The AAUT-ML has been evaluated using three datasets data mining (DM), operating system (OS), and data base (DB), and compared with the existing models, i.e., YAKE, term frequency-inverse document frequency (TF-IDF) and text-R. The results show better outcomes in terms of precision, recall, and macro-averaged F1-score. This work presents a novel method for identifying complex semantics using unstructured data.

Many Web information retrieval methods are related to special Web sites, for example, the method based on extracting rules and the one based on training page samples. These methods can do well in a Web site but fail in the others without adding new rules or inputting new training pages manually. Furthermore, if the template of the Web site is changed, it has to redesign the extracting rules or re-inputting the training pages. It is hard to be maintained and used to extract information from large number of different Web sites. In the paper, there is a new method that is based on the keywords of a certain topic, instead of rules and training pages. Experimental evaluation on a large of Web pages from different Web sites indicates that this method correctly and automatically extracts the information ignoring which Web sites the pages come from. This method has been applied to the system of intelligent searching and mining of electronic business successfully.

Classifying or predicting occupational incidents using both structured and unstructured (text) data are an unexplored area of research. Unstructured texts, i.e., incident narratives are often unutilized or underutilized. Besides the explicit information, there exist a large amount of hidden information present in a dataset, which cannot be explored by the traditional machine learning (ML) algorithms. There is a scarcity of studies that reveal the use of deep neural networks (DNNs) in the domain of incident prediction, and its parameter optimization for achieving better prediction power. To address these issues, initially, key terms are extracted from the unstructured texts using LDA-based topic modeling. Then, these key terms are added with the predictor categories to form the feature vector, which is further processed for noise reduction and fed to the adaptive moment estimation (ADAM)-based DNN (i.e., ADNN) for classification, as ADAM is superior to GD, SGD, and RMSProp. To evaluate the effectiveness of our proposed method, a comparative study has been conducted using some state-of-the-arts on five benchmark datasets. Moreover, a case study of an integrated steel plant in India has been demonstrated for the validation of the proposed model. Experimental results reveal that ADNN produces superior performance than others in terms of accuracy. Therefore, the present study offers a robust methodological guide that enables us to handle the issues of unstructured data and hidden information for developing a predictive model.

This study aims to develop high-performing Machine Learning and Deep Learning models in predicting hospital length of stay (LOS) while enhancing interpretability. We compare performance and interpretability of models trained only on structured tabular data with models trained only on unstructured clinical text data, and on mixed data. The structured data was used to train fourteen classical Machine Learning models including advanced ensemble trees, neural networks and k-nearest neighbors. The unstructured data was used to fine-tune a pre-trained Bio Clinical BERT Transformer Deep Learning model. The structured and unstructured data were then merged into a tabular dataset after vectorization of the clinical text and a dimensional reduction through Latent Dirichlet Allocation. The study used the free and publicly available Medical Information Mart for Intensive Care (MIMIC) III database, on the open AutoML Library AutoGluon. Performance is evaluated with respect to two types of random classifiers, used as baselines. The best model from structured data demonstrates high performance (ROC AUC = 0.944, PRC AUC = 0.655) with limited interpretability, where the most important predictors of prolonged LOS are the level of blood urea nitrogen and of platelets. The Transformer model displays a good but lower performance (ROC AUC = 0.842, PRC AUC = 0.375) with a richer array of interpretability by providing more specific in-hospital factors including procedures, conditions, and medical history. The best model trained on mixed data satisfies both a high level of performance (ROC AUC = 0.963, PRC AUC = 0.746) and a much larger scope in interpretability including pathologies of the intestine, the colon, and the blood; infectious diseases, respiratory problems, procedures involving sedation and intubation, and vascular surgery. Our results outperform most of the state-of-the-art models in LOS prediction both in terms of performance and of interpretability. Data fusion between structured and unstructured text data may significantly improve performance and interpretability.

Web sites have become essential tools in the dissemination of content in digital format, especially in organisations where the generation, management and distribution of information and knowledge are among the major activities. This should also be true in agricultural research organisations, where information and knowledge are the major byproducts of agricultural research. This point is supported by the results of a content survey of 28 Web sites of international agricultural organisations, eight Web sites of regional-based agricultural research organisations in Africa and 21 Web sites of organisations involved in agricultural research in the Southern Africa Development Community (SADC). Overall, international agricultural organisations are way above agricultural research organisations in Africa and the SADC region in the use of Web sites in providing access to information and knowledge resources in digital formats. Agricultural organisations in Africa and the SADC region are generally behind in implementing innovative use of the Web. They need to be guided in the development and provision of digital information services using the Internet and Web technologies.

Context. Currently, there are a lot of approaches that are used for textual search. Nowadays, methods such as pattern-matching and optical character recognition are highly used for retrieving preferred information from documents with proven effectiveness. However, they work with a common or predictive document structure, while unstructured documents are neglected. The problem – is automating the textual search in documents with unstructured content. The object of the study was to develop a method and implement it into an efficient model for searching the content in unstructured textual information. Objective. The goal of the work is the implementation of a rule-based textual search method and a model for seeking and retrieving information from documents with unstructured text content. Method. To achieve the purpose of the research, the method of rule-based textual search in heterogenous content was developed and applied in the appropriately designed model. It is based on natural language processing that has been improved in recent years along with a new generative artificial intelligence becoming more available. Results. The method has been implemented in a designed model that represents a pattern or a framework of unstructured textual search for software engineers. The application programming interface has been implemented. Conclusions. The conducted experiments have confirmed the proposed software’s operability and allow recommendations for use in practice for solving the problems of textual search in unstructured documents. The prospects for further research may include the improvement of the performance using multithreading or parallelization for large textual documents along with the optimization approaches to minimize the impact of OpenAI application programming interface content processing limitations. Furthermore, additional investigation might incorporate extending the area of imperative variables usage in programming and software development.

Machine-readable semantic knowledge in the form of taxonomies (i.e., a collection of is-a edges) has proved to be beneficial in an array of NLP tasks including inference, textual entailment, question answering and information extraction. Such widespread utility of taxonomies has led to multiple large-scale manual efforts towards taxonomy induction such as WordNet and Cyc. However, manual construction of taxonomies is time-intensive, and usually, requires substantial annotation efforts by domain experts. Furthermore, the resulting taxonomies suffer from low coverage and are unavailable for specific domains or languages. Therefore, in recent years, there has been a growing body of work, which aims to induce taxonomies automatically, either from semi-structured knowledge resources or unstructured text. In this thesis, we focus on the task of automated taxonomy induction under a variety of different settings. We first focus on taxonomy induction from the largest semi-structured knowledge resource, i.e., Wikipedia. More specifically, we introduce a set of novel heuristics aimed towards inducing a large-scale taxonomy from the English Wikipedia categories network. We also propose a novel comprehensive path-based evaluation framework for taxonomies. Taxonomy induced using our approach significantly outperforms the state of the art across both edge-based as well as path-based evaluation metrics. Moreover, our experiments also demonstrate that high accuracy in edge-based evaluation metrics does not always translate to high-accuracy in path-based evaluation metrics. Subsequently, we propose a novel approach, which leverages the interlanguage links of Wikipedia to induce taxonomies in other Wikipedia languages. Compared to the state of the art, our approach is simpler, more principled, and results in taxonomies that are significantly more accurate across both edge-based and path-based evaluation metrics. In the second part of this thesis, we focus on taxonomy induction from unstructured text. We propose a novel approach towards taxonomy induction from an input vocabulary of seed terms. Unlike previous approaches, which extract singular hypernym edges for terms, our approach utilizes a novel probabilistic framework to extract long-range hypernym subsequences. Taxonomy induction from the extracted subsequences is cast as an instance of the minimum-cost flow optimization problem on a carefully-designed flow network. Through experiments, we demonstrate that our approach outperforms the state of the art across four languages. We also show that our approach is robust to the presence of noise in the input vocabulary. Our approach facilitates the relaxation of many simplifying assumptions, which were employed by previous taxonomy induction approaches. As a result, our work serves to automate the process of taxonomy induction from unstructured text in the true sense. Finally, we introduce a task of discovering and generalizing lexicalized templates from the titles of Wikipedia entities. The experimental results on this task demonstrate that taxonomies, which perform better on our proposed path-based metrics, result in a more accurate set of generalizations for a given set of entities. In summary, this thesis proposes new approaches towards automated taxonomy induction. It improves upon the state of the art in a variety of different settings. It also relaxes many simplifying assumptions that limited the applicability of prior approaches.

Classroom lessons are conducted through interactions between teachers and students. Teacher factor, in particular, has recently been emphasized.BR In this research, ‘I’, as a history teacher, reflect on my own classes through teaching diary and analyze how teachers exploit their knowledges and how students respond in return, which would be conducive to effective teaching plans. This study was performed by the teacher herself and utilized the reflective techniques through self-reflective teaching diary, targeting second and third grade of a middle school in Chungnam, and 46 periods of history class have been meticulously observed and researched from March 1 to November 29, 2013.BR The research results are as follows. First, as the resource of the teacher knowledge, the researcher exploits her own historical knowledges, together with information on teaching materials, her experiences back in secondary school and content knowledges acquired in college, other teachers’ web sites and demonstration classes. Second, the researcher develops the lesson in order to have students understand the historical context, aiming to facilitate students’ understanding of the present through past fact. Third, it appears that a great part of classes can be developed by story telling by analyzing teaching diary.BR Finally, teaching and learning activities in accordance with the students’ understanding of history are analyzed in this research. One of the difficulties the researcher frequently faces in conducting history class, especially when teaching concepts and terms, is that she is not fully capable of figuring out how well students understand. In this sense, reflections through teaching diary enable teachers to look back on their own classes, raise awareness of the problems occurred, and help them develop their specialties.

The World Wide Web rapidly guides the world into a newly amazing electronic world, where everyone can publish anything in electronic form and extract almost all the information. Extraction of information from semi structured or unstructured documents, such as web pages, is a useful yet complex task. Data extraction, which is important for many applications, extracts the records from the HTML files automatically. Ontologies can achieve a high degree of accuracy in data extraction. We analyze method for data extraction OBDE (Ontology‐Based Data Extraction), which automatically extracts the query result records from the web with the help of agents. OBDE first constructs an ontology for a domain according to information matching between the query interfaces and query result pages from different web sites within the same domain. Then, the constructed domain ontology is used during data extraction to identify the query result section in a query result page and to align and label the data values in the extracted records. The ontology‐assisted data extraction method is fully automatic and overcomes many of the deficiencies of current automatic data extraction methods.

BackgroundBERT models have seen widespread use on unstructured text within the clinical domain. However, little to no research has been conducted into classifying unstructured clinical notes on the basis of patient lifestyle indicators, especially in Dutch. This article aims to test the feasibility of deep BERT models on the task of patient lifestyle classification, as well as introducing an experimental framework that is easily reproducible in future research.MethodsThis study makes use of unstructured general patient text data from HagaZiekenhuis, a large hospital in The Netherlands. Over 148 000 notes were provided to us, which were each automatically labelled on the basis of the respective patients’ smoking, alcohol usage and drug usage statuses. In this paper we test feasibility of automatically assigning labels, and justify it using hand-labelled input. Ultimately, we compare macro F1-scores of string matching, SGD and several BERT models on the task of classifying smoking, alcohol and drug usage. We test Dutch BERT models and English models with translated input.ResultsWe find that our further pre-trained MedRoBERTa.nl-HAGA model outperformed every other model on smoking (0.93) and drug usage (0.77). Interestingly, our ClinicalBERT model that was merely fine-tuned on translated text performed best on the alcohol task (0.80). In t-SNE visualisations, we show our MedRoBERTa.nl-HAGA model is the best model to differentiate between classes in the embedding space, explaining its superior classification performance.ConclusionsWe suggest MedRoBERTa.nl-HAGA to be used as a baseline in future research on Dutch free text patient lifestyle classification. We furthermore strongly suggest further exploring the application of translation to input text in non-English clinical BERT research, as we only translated a subset of the full set and yet achieved very promising results.

This paper presents our algorithmic approach for information and relation extraction from unstructured texts (such as from eBook sections or webpages), performing other useful analytics on the text, and automatically generating a semantically meaningful structure (RDF schema). Our algorithmic formulation parses the unstructured text from eBooks and identifies key concepts described in the eBook along with relationship between the concepts. The extracted information is then used for four purposes: (a) for generating some computed metadata about the text source (such as readability of an eBook), (b) generate a concept profile for each distinct part of text, (c) identifying and plotting relationship between key concepts described in the text, and (d) to generate RDF representation for the text source. We have done our experiments on eBook texts from Computer Science domain; however, the approach can be applied to work on different forms of text in other domains as well. The results are not only useful for concept based tagging and navigation of unstructured text documents (such as eBook) but can also be used to design a comprehensive and sophisticated learning recommendation system.

The task of insights extraction from unstructured text poses significant challenges for big data analytics because it contains subjective intentions, different contextual perspectives, and information about the surrounding real world. The technical and conceptual complexities of unstructured text degrade its usability for analytics. Unlike structured data, the existing literature lacks solutions to address the usability of unstructured text big data. A usability enhancement model has been developed to address this research gap, incorporating various usability dimensions, determinants, and rules as key components. This paper adopted Delphi technique to validate the usability enhancement model to ensure its correctness, confidentiality, and reliability. The primary goal of model validation is to assess the external validity and suitability of the model through domain experts and professionals. Therefore, the subject matter experts of industry and academia from different countries were invited to this Delphi, which provides more reliable and extensive opinions. A multistep iterative process of Knowledge Resource Nomination Worksheet (KRNW) has been adopted for expert identification and selection. Average Percent of Majority Opinions (APMO) method has been used to produce the cut-off rate to determine the consensus achievement. The consensus was not achieved after the first round of Delphi, whereas APMO cut-off rate was 70.9%. The model has been improved based on the opinions of 10 subject matter experts. After second round, the analysis has shown majority agreement for the revised model and consensus achievement for all improvements that validate the improved usability enhancement model. The final proposed model provides a systematic and structured approach to enhance the usability of unstructured text big data. The outcome of the research is significant for researchers and data analysts.

Spatial objects often come with textual information, such as Points of Interest (POIs) with their descriptions, which are referred to as geo-textual data. To retrieve such data, spatial keyword queries that take into account both spatial proximity and textual relevance have been extensively studied. Existing indexes designed for spatial keyword queries are mostly built based on the geo-textual data without considering the distribution of queries already received. However, previous studies have shown that utilizing the known query distribution can improve the index structure for future query processing. In this paper, we propose WISK, a learned index for spatial keyword queries, which self-adapts for optimizing querying costs given a query workload. One key challenge is how to utilize both structured spatial attributes and unstructured textual information during learning the index. We first divide the data objects into partitions, aiming to minimize the processing costs of the given query workload. We prove the NP-hardness of the partitioning problem and propose a machine learning model to find the optimal partitions. Then, to achieve more pruning power, we build a hierarchical structure based on the generated partitions in a bottom-up manner with a reinforcement learning-based approach. We conduct extensive experiments on real-world datasets and query workloads with various distributions, and the results show that WISK outperforms all competitors, achieving up to 8× speedup in querying time with comparable storage overhead.

Many online labor markets face challenges in matching service providers with employers because of the high uncertainty in online environments compared to traditional ones. While prior studies have mostly focused on the uncertainty that employers face with service providers, our study explores the uncertainty service providers face when selecting projects on which to bid. Specifically, our study examines the role of project description in the Call for Bids (CFB) that describes the services requested by employers, in shaping service providers’ bidding behavior and the matching between service providers and employers. Drawing upon transaction cost economics, we examine three key dimensions of project descriptions: (a) c odifiability (provisions of detailed procedures), (b) flexibility (provisions of renegotiation opportunities), and (c) provisions of outcome standards . We first use machine learning algorithms to quantify the three key dimensions based on unstructured textual information. Second, we use Bayesian mediation analysis and a randomized experiment to examine the role of these three key dimensions in matching efficiency and to explore the mediating role of three key bid attributes (i.e., number of bids, average price of bids, average quality of bids). Findings reveal: First , codifiability is positively associated with matching efficiency by reducing the number of bids received by employers. Second , flexibility is negatively associated with matching efficiency by raising the average price of bids. Third , outcome standards are positively linked to matching efficiency by reducing the number, albeit improving the average quality, of bids. By focusing on the project description and analyzing the textual content of CFB, our study contributes to the literature by enhancing the matching efficiency of online labor markets and facilitating the contracting of IT services in online labor markets.

This paper proposes a set of approaches for the semantic search of mobile applications, based on their name and on the unstructured textual information contained in their description. The proposed approaches make use of word-level, character-level, and contextual word-embeddings that have been trained or fine-tuned using a dataset of about 500 thousand mobile apps, collected in the scope of this work. The proposed approaches have been evaluated using a public dataset that includes information about 43 thousand applications, and 56 manually annotated non-exact queries. Our results show that both character-level embeddings trained on our data, and fine-tuned RoBERTa models surpass the performance of the other existing retrieval strategies reported in the literature.