Text Mining Algorithms Research Articles

Assessing text mining algorithm outcomes

ABSTRACT There is a surge in the development of decision-oriented analysis tools intended to extract actionable information from text. These tools integrate various text-mining methods that were performance tested in a manner that was often biased toward the new system. Those tests primarily utilised descriptive measurement criteria and test datasets that are inconsistent with most business corpora. We propose and test a user-oriented judgment approach that allows testing under controlled customer-oriented corpora and generates effect size measures. To illustrate the approach, customer relations data was analysed by latent semantic analysis and latent Dirichlet analysis with results evaluated by prospective business analysts. Reporting includes comparisons of results with published literature. While the research centres on the context-region text-mining systems, literature comparisons include word-embedding methods. The analysis concludes that none of the systems reviewed possess a repeatable statistical advantage over the others. Instead, distribution attributes, algorithm configuration, and the evaluation task drive results.

Comparative Pathway Integrator: A Framework of Meta-Analytic Integration of Multiple Transcriptomic Studies for Consensual and Differential Pathway Analysis.

Pathway enrichment analysis provides a knowledge-driven approach to interpret differentially expressed genes associated with disease status. Many tools have been developed to analyze a single study. However, when multiple studies of different conditions are jointly analyzed, novel integrative tools are needed. In addition, pathway redundancy introduced by combining multiple public pathway databases hinders interpretation and knowledge discovery. We present a meta-analytic integration tool, Comparative Pathway Integrator (CPI), to address these issues using adaptively weighted Fisher’s method to discover consensual and differential enrichment patterns, a tight clustering algorithm to reduce pathway redundancy, and a text mining algorithm to assist interpretation of the pathway clusters. We applied CPI to jointly analyze six psychiatric disorder transcriptomic studies to demonstrate its effectiveness, and found functions confirmed by previous biological studies as well as novel enrichment patterns. CPI’s R package is accessible online on Github metaOmics/MetaPath.

A predictive model of tourist destinations based on tourists' comments and interests using text analytics

Data provided by tourists always benefit tourism managers and help them offer customized services, products and destinations to future travelers. This research investigates the effect of interests on Iranian outbound tourists, especially their selection of a destination and then, using text and data mining algorithms, it introduces a model to predict tourists' destinations based on their interests and travel backgrounds. In the current study, a dataset of 244,980 travels, consisting of 6661 people, was extracted from social media to discover the relationship between tourists' interests and travel destinations. Hence, it represents a model that is created using data and text mining from travel agencies to design their marketing plans by offering and advertising destinations to travelers with specific interest categories. The model has also shown promising accuracy and interesting results for the future tourist destination data and text analysis.

Text Mining Method in the Field of Health

Objective: Text mining which digitalizes textual data and enables them to be applied for text mining algorithms has very important place in the today’s world. The aim of this study was to introduce the text mining method and to show its application on a subject in the field of health. Methods: The text mining method was applied to the individual documents obtained from the most commonly used Pubmed database and then the merged documents under two different titles as “human-and-cancer” and “mouse-and-cancer” through the Knime program, and the document classification was made using K nearest neighbor (K-NN) algorithm. Results: The prominent words were “cell” and “cancer” in tag cloud graphs. In both documents, the words such as “cell”, “cancer”, “tumor”, “patient”, whose frequency values were high, were observed to be high rates in the analysis performed after the data was merged. It was found that 255 of 600 test documents belonged to the human-and-cancer class and the remaining belonged to the mouse-and-cancer class, and the accuracy classification was 56.6% for the human-and-cancer-documents and 62.6% for the mouse-and-cancer-documents according to the F-criteria. It was determined that the document classification estimation by the K-NN algorithm was relatively successful with a rate of 59.8% however Cohen’s kappa value was 19.7%, meaning that the fit was of slight level. Conclusion: It was recommended to use the text mining method and to generalize its use in order to obtain information quickly and reliably in the health field where there were numerous digital and printed documents.

TeamTat: a collaborative text annotation tool.

Manually annotated data is key to developing text-mining and information-extraction algorithms. However, human annotation requires considerable time, effort and expertise. Given the rapid growth of biomedical literature, it is paramount to build tools that facilitate speed and maintain expert quality. While existing text annotation tools may provide user-friendly interfaces to domain experts, limited support is available for figure display, project management, and multi-user team annotation. In response, we developed TeamTat (https://www.teamtat.org), a web-based annotation tool (local setup available), equipped to manage team annotation projects engagingly and efficiently. TeamTat is a novel tool for managing multi-user, multi-label document annotation, reflecting the entire production life cycle. Project managers can specify annotation schema for entities and relations and select annotator(s) and distribute documents anonymously to prevent bias. Document input format can be plain text, PDF or BioC(uploaded locally or automatically retrieved from PubMed/PMC), and output format is BioC with inline annotations. TeamTat displays figures from the full text for the annotator's convenience. Multiple users can work on the same document independently in their workspaces, and the team manager can track task completion. TeamTat provides corpus quality assessment via inter-annotator agreement statistics, and a user-friendly interface convenient for annotation review and inter-annotator disagreement resolution to improve corpus quality.

Credit Card Fraud Detection Performance Improvement using Advanced Super Gradient Boosting Algorithm

Credit card fraud introduces to the physical loss of a credit card or the destruction of sensitive credit card data. Several text mining procedures can be used for disclosure. This investigation reveals several algorithms that can be used to analyze transactions as a fraud or as a real background. This paper represents the possibility of fraudulent transactions in the prevalence and meaning of credit card usage also, Credit card fraud data collection was used in the investigation. Since the dataset was largely unbalanced, SMOTE (Synthetic Minority oversampling Technique) is applying for an overdose. In addition, jobs selected, and the data set divided into two parts, training data and test data. In this paper, The Advanced Super Gradient Boostingbased Text mining Algorithm (ASGB) suggested to detect the fraud transaction in Credit card transactions. ASGB is a Decision-Tree-Based Ensemble Text mining algorithm that utilizes a gradient boosting framework. In forecast difficulties, including unstructured data (Images, Text, etc.), artificial neural networks tend to exceed all other algorithms or structures. The proposed algorithms used in the experiment were the Hidden Markov Model, Random Forest, Gradient Boosting, and Enhanced Hidden Markov Model. The Experimental Results show that proposed algorithms, a well-tuned ASGB classifier outperforms all of them. And it presents better Precision is 99.1%, and Recall is 99.8%, F-measure is 99.5%.

Research trends on big data domain using text mining algorithms

Abstract Most of the theories have considered big data as an interesting subject in the information technology domain. Big data is a term for describing huge databases that traditional methods in data processing suffer from analyzing them. Recognizing and clustering emerging topics in this area will help researchers whose aim is to work on this interesting subject. Text mining and social network analysis algorithms are utilized for identifying the emerging trends for big data domain. In this study, at first, we gathered the whole papers that are relevant to big data domain and then the word co-occurrence network was created based on the extracted keywords. Then the best clusters were identified and the relationship between keywords was recognized by the association rules technique. In conclusion, some suggestions were mentioned for future studies.

Focal points for sustainable development strategies—Text mining-based comparative analysis of voluntary national reviews

Countries have to work out and follow tailored strategies for the achievement of their Sustainable Development Goals. At the end of 2018, more than 100 voluntary national reviews were published. The reviews are transformed by text mining algorithms into networks of keywords to identify country-specific thematic areas of the strategies and cluster countries that face similar problems and follow similar development strategies. The analysis of the 75 VNRs has shown that SDG5 (gender equality) is the most discussed goal worldwide, as it is discussed in 77% of the analysed Voluntary National Reviews. The SDG8 (decent work and economic growth) is the second most studied goal, With 76 %, while the SDG1 (no poverty) is the least focused goal, it is mentioned only in 48 % of documents and the SDG10 (reduced inequalities) in 49 %. The results demonstrate that the proposed benchmark tool is capable of highlighting what kind of activities can make significant contributions to achieve sustainable developments.

An Assessment of Idea Emergence in Subject-Matter Collaborative Learning

Idea emergence is critical in learning as knowledge creation. Although recent advancements make it possible to detect emergent ideas and evaluate how students engage in knowledge creation in collaborative learning contexts, the relationship between the learning processes and final learning outcomes has not been well studied. One compound factor is how much students engage in their study topic during collaboration. In this study, therefore, we propose a new procedure for evaluating idea emergence in the context of jigsaw instruction by combining a socio-semantic network analysis of discourse and a text-mining algorithm, “the term-frequency.” This procedure was used to evaluate how high-school learners engaged in their social process of knowledge creation as well as how much they discuss their study topics, the human immune system. Results showed that the weight of priority on a study topic did not significantly differ in both high and low learning-outcome groups, but high learning-outcome groups were more engaged in sharing and discussing ideas in the early stage of their collaboration. It is suggested that students’ recognition of the study topic was not different depending on the levels of their learning outcomes in a well-structured collaborative learning context such as jigsaw instruction. However, what matters is how students discuss their ideas through collaborative discourse.

Abstract P161: A Learning Healthcare System Improves Cardiovascular Risk Management: Results From the Utrecht Cardiovascular Cohort Initiative

Introduction: Cardiovascular disease (CVD) is a good example of a high complex, multi-morbid, chronic condition. This asks for multidisciplinary and transmural collaboration and communication between caretakers, forming a care continuum. Research is often conducted outside routine practice, namely in randomized controlled trials and dedicated cohorts with strict in- and exclusion criteria. Since our real world patient does not fit these criteria, generated evidence does not always translate to practice. Yet, we can learn from each patient. Within a learning healthcare system (LHS), data collection in routine daily practice is used as input for analysis , interpretation and actionable feedback . CVD management (CVRM) guidelines provide various recommendations for diagnosis, modification and referral from primary to secondary or tertiary care for the evaluation of a CV condition. Starting with uniform structured collection of the CVD risk profile in the our hospital in all patients, we developed a learning health care system focused on CVD risk management. Hypothesis: We hypothesize that a learning health care system improves CVD risk management. Methods: Embedded in routine clinical practice, the Utrecht Cardiovascular Cohort (UCC) supports structured and uniform assessment and registration of guideline based CVD risk profile in the electronic health record across all departments. We studied, with data from 2800 participants, data collection (uniform and linkage); data extraction ((un)structured); data analysis (missings); and actionable feedback (computer decision support system). Results: Establishment of UCC resulted in more complete CVD risk profile: increase in risk factor completeness ranged from 8% (eGFR) to 50% (HbA1c). Linking UCC patient data to general practitioner records, showed considerable gaps in transmural CVD communication: from GP to specialist and from specialist to GP. To address missing information in electronic health records we developed a text mining algorithm to detect smoking status: algorithm showed a 99% yield, a positive and negative predictive value of 65% and 95%. We developed strategies to handle missing data for risk prediction on an individual level in a ‘live’ setting: joint conditional model led to a lower mean squared error (the lower, the more precise estimates). We developed a CVD risk management computerized decision support system embedded in our electronic health record, filled with extracted information and incorporating all lessons learned from reviews, clinician expert opinions, and patient expert opinions. This dashboard, favorably received by physicians, shows in the consulting room the CVD risk profile, potential treatment effects in one glance and helps with shared decision making. Conclusions: Our learning health care approach shows improvement in some areas and provides actionable leads for further improvement.

Developing a Surgical Site Infection Surveillance System Based on Hospital Unstructured Clinical Notes and Text Mining

Background: Electronic surveillance using clinical and administrative data from multiple sources has been reported as a tool for surveillance of surgical site infections (SSIs), but experiences are limited. In this study, we aimed to assess the accuracy of a text-searching algorithm to detect SSIs in children based on the application of regular expressions of unstructured clinical notes collected through different information systems. Methods: We developed an information system data warehouse that integrates data provided by electronic health and administrative records for patients who underwent surgical procedures in index weeks when active SSIs surveillances was conducted. To capture whether the patient developed an SSI, we developed a customized application to analyze clinical notes and code descriptions applying a pattern-matching algorithm based on regular expressions. We described the SSI cases detected by the active surveillance and the text-searching algorithm. To assess the accuracy in identifying the SSIs through the two methods, we adopted a reference standard that calculated the total number of SSIs as those detected by active surveillance plus those derived by the text-searching algorithm that was missed by active surveillance. Results: Compared with the total number of SSIs used as a reference standard, both methods had a specificity of 100%, a positive predictive value of 100%, and a negative predictive value >99.5%. Sensitivity was 70% for the text-mining algorithm and 60% for the active surveillance. Accuracy was >99% with both methods. The kappa value was 0.46. Conclusions: Compared with conventional surveillance of SSIs, a text-searching algorithm is a valid tool for case finding that has the potential to reduce drastically the workload of conventional surveillance, which involved direct contact with all families.

Forecasting health of complex IT systems using system log data

Predicting the health of digital infrastructure is a vital issue to minimize downtime to maintain a high service level. This research work has applied predictive analytics to forecast future health of complex-IT (Information Technology) based infrastructure. Every subset of a complex IT infrastructure is at the level of a single machine, that tracks the run-time status of the system and generates electronic messages, error events, and further manual massages as ticket logs. This research has suggested a 3-step method to build a novel predictive analytics model using text mining algorithm for extracting features from the log data. Then, it provides a model for selecting the critical devices in the system and predicting their failure. The final step suggests a forecasting model to predict the health of infrastructure for a given timestamp. The models in the second step of this integrated approach are built using algorithms such as association rules and rank based algorithm. The time-series model is built using a machine learning methods (ANN and SVR). This approach can be readily applied to many other types of information technology-based medical, banking, energy infrastructure, and other applications.

Natural Language Processing in Dutch FreeText Radiology Reports: Challenges in a Small Language Area Staging Pulmonary Oncology.

Reports are the standard way of communication between the radiologist and the referring clinician. Efforts are made to improve this communication by, for instance, introducing standardization and structured reporting. Natural Language Processing (NLP) is another promising tool which can improve and enhance the radiological report by processing free text. NLP as such adds structure to the report and exposes the information, which in turn can be used for further analysis. This paper describes pre-processing and processing steps and highlights important challenges to overcome in order to successfully implement a free text mining algorithm using NLP tools and machine learning in a small language area, like Dutch. A rule-based algorithm was constructed to classify T-stage of pulmonary oncology from the original free text radiological report, based on the items tumor size, presence and involvement according to the 8th TNM classification system. PyContextNLP, spaCy and regular expressions were used as tools to extract the correct information and process the free text. Overall accuracy of the algorithm for evaluating T-stage was 0,83 in the training set and 0,87 in the validation set, which shows that the approach in this pilot study is promising. Future research with larger datasets and external validation is needed to be able to introduce more machine learning approaches and perhaps to reduce required input efforts of domain-specific knowledge. However, a hybrid NLP approach will probably achieve the best results.

Implementation of Integrated Bayes Formula and Support Vector Machine for Analysing Airline’s Passengers Review

Nowadays, the utilization of Internet of Things (IoT) is commonly used in the tourism industry, including aviation, where passengers of flight services can rate their satisfaction levels towards the product and service they use by writing their reviews in the form of text-based data on many popular websites. These passenger reviews are collections of potential big data and can be analyzed in order to extract meaningful informations. Some text mining algorithms are already in common use, including the Bayes formula and Support Vector Machine methods. This research proposes an implementation of the Bayes and SVM methods where these algorithms will operate independently yet integrated with other modules such as input data, text pre-processing and shows output result concisely in one single information system. The proposed system was successfully delivered 1000 documents of passenger reviews as input data, then after implemented the pre-processing method, the Bayes formula was used to classify the document reviews into 5 categories, including plane condition, flight comfort, staff service, food and entertainment, and price. While simultanously, the positive and negative sentiment contained in the review document was analyzed with SVM method and shows the accuracy score of 83.6% for a training to testing set ratio of 50:50, while 82.75% accuracy for the 60:40 ratio, and 83.3% accuracy for the 70:30 ratio. This research shows that two different text mining algorithms can be implemented simultaneously in a effective and efficient way, while still providing an accurate and satisfying performance results in one integrated information system.

Distinguishing True and Fake News by Using Text Mining and Machine Learning Algorithm

With recent advancements in social media and technology as a whole, online news sources have increased. Therefore there has been a higher demand of people wanting a convenient way to find recent, relevant and updated online news articles and posts from social media platforms. In the current status quo, many people feel comfortable with their main source of news being social media articles. Unfortunately, receiving news via social media platforms and unverified online sites has aroused many problems, one of which being fake news (news which contain incorrect or biased facts and statements). Many individuals all around the world are vulnerable and subject to fake news and becoming victims of propaganda and/or being misinformed. To solve this world-wide complication, we used word preprocessing skills to digest the content of articles, and used several mathematical vectors to pinpoint the legitimacy of a news article. To establish an accurate system, words used in examples of fake news and real news were collected using Python. Verifying fake and real news is an important process that all news should go through as it can result in immense consequences. Data on real news and fake news were collected from Kaggle. We had the conclusion that the trained machine learning algorithms showed high accuracy of distinguishing which indicates our research was successful.

Large-scale semantic exploration of scientific literature using topic-based hashing algorithms

Searching for similar documents and exploring major themes covered across groups of documents are common activities when browsing collections of scientific papers. This manual knowledge-intensive task can become less tedious and even lead to unexpected relevant findings if unsupervised algorithms are applied to help researchers. Most text mining algorithms represent documents in a common feature space that abstract them away from the specific sequence of words used in them. Probabilistic Topic Models reduce that feature space by annotating documents with thematic information. Over this low-dimensional latent space some locality-sensitive hashing algorithms have been proposed to perform document similarity search. However, thematic information gets hidden behind hash codes, preventing thematic exploration and limiting the explanatory capability of topics to justify content-based similarities. This paper presents a novel hashing algorithm based on approximate nearest-neighbor techniques that uses hierarchical sets of topics as hash codes. It not only performs efficient similarity searches, but also allows extending those queries with thematic restrictions explaining the similarity score from the most relevant topics. Extensive evaluations on both scientific and industrial text datasets validate the proposed algorithm in terms of accuracy and efficiency.

Latent Feature Word Representations to Enhance Topic Models for Text Mining Algorithms

Dealing with large number of textual documents needs proven models that leverage the efficiency in processing. Text mining needs such models to have meaningful approaches to extract latent features from document collection. Latent Dirichlet allocation (LDA) is one such probabilistic generative process model that helps in representing document collections in a systematic approach. In many text mining applications LDA is useful as it supports many models. One such model is known as Topic Model. However, topic models LDA needs to be improved in order to exploit latent feature vector representations of words trained on large corpora to improve word-topic mapping learnt on smaller corpus. With respect to document clustering and document classification, it is essential to have a novel topic models to improve performance. In this paper, an improved topic model is proposed and implemented using LDA which exploits the benefits of Word2Vec tool to have pre-trained word vectors so as to achieve the desired enhancement. A prototype application is built to demonstrate the proof of the concept with text mining operations like document clustering.

Applied Text-Mining Algorithms for Stock Price Prediction Based on Financial News Articles

This article includes a developed model and well-defined process that one should undertake in order to contribute in the prediction of the potential stock price fluctuation solely based on financial news from relevant sources. We are providing background information on this topic adding the role of text mining in general, furthermore supporting the idea with the study of relevant research articles to narrow the focus on the problemwe are researching.Our proposedmodel relies on existing text-mining techniques used for sentiment analysis, combinedwith historical data from relevant news sources as well as stock data. In confirming the model, after the experiment we have provided the results of the simulation, which are opening the ground for further explorations in this sensitive area of prediction.

Revisiting three decades of Biology and Philosophy: a computational topic-modeling perspective

Though only established as a discipline since the 1970s, philosophy of biology has already triggered investigations about its own history (e.g., Grene and Depew in The philosophy of biology: an episodic history, Cambridge University Press, Cambridge, 2004. https://doi.org/10.1017/CBO9780511819018; Hull, in: Ruse (ed) The Oxford handbook of philosophy of biology, Oxford University Press, New York, pp 11–33, 2008). When it comes to assessing the road since travelled—the research questions that have been pursued—manuals and ontologies also offer specific viewpoints, highlighting dedicated domains of inquiry and select work. In this article, we propose to approach the history of the philosophy of biology with a complementary data-driven perspective that makes use of statistical algorithms applied to the complete full-text corpus of one major journal of the field—Biology and Philosophy—from its launch in 1986 up until 2017. By running text-mining and topic-modeling algorithms, we identified 67 key research topics that span across these 32 years. We also investigated the evolution of these topics over time and their fluctuating significance in the journal articles. Our results concur with known episodes or traits of the discipline—for instance, the significance of evolution-related topics or the decrease of articles with a marked historical dimension—but also highlight a diversity of topics that is much richer than what is usually acknowledged.

ANALISA TESTIMONIAL DENGAN MENGGUNAKAN ALGORITMA TEXT MINING DAN TERM FREQUENCY- INVERSE DOCUMENT FREQUENCE (TF-IDF) PADA TOKO ALLMEEART

E-commerce or often referred to as an online shop is the latest trend of the community in carrying out shopping activities, first before the rise of e-commerce companies like today the community to meet their needs still rely on distros around the customer lives, or to a shopping place but now it has switch to shoop online. The advantages offered by online shoop are the relatively low prices, no need to shop locations, and guarantee goods, it has an impact on retail shops that are increasingly lonely. Testimonials are one of the techniques carried out to convince customers to shop at e-commerce they have, testimonials are the responses of buyers for their experience of shopping in an e-commerce application starting from the payment process until the goods are received, the more positive experiences conveyed in the testimonials, the customer who have not shopped on an e-commerce application will be more convinced to shop. Testimonials on an e-commerce application are not always positive, there are times when testimonials are delivered by negative buyers. The customer's problem is the unavailability of percentages or information on the number of buyers with positive and negative shopping experiences because in general testimonials are only delivered in the form of a list.Keywords: Testimonial Analysis, Text Mining Algorithm, Term Frequency-Inverse Document Frequency (TF-IDF)