Trend Analysis in Machine Learning Research Using Text Mining

This study aims to uncover recent trends and themes in text mining research, focusing on online news and annual report data sources. These sources are rich in content, real-world context, and domain-specific information, making them crucial for text mining. The primary research questions guiding this study are: 1) What are the emerging trends in text mining research applied to online news and annual reports? 2) What are the themes generated based on systematic literature review and text mining technique? To address this question, the study systematically reviews a large number of related studies using the PRISMA review protocol and text mining techniques from the SCOPUS and Web of Science databases. After thorough evaluation, 34 selected articles were analyzed. The PRISMA review protocol ensures transparency and completeness in reporting the review process through its standardized approach for systematic reviews. Additionally, this systematic review explores advancements in text mining techniques such as document clustering and topic modeling, which have facilitated the identification and extraction of relevant evidence from vast amounts of textual data. The study’s findings identified four primary themes (text mining/text analytics, machine learning, deep learning, ensemble methods) with 19 sub-themes related to each theme’s methodology when applying the PRISMA protocol. By utilizing a text mining techniques, five topics were uncovered based on article keywords (text mining, text analytics, machine learning, deep learning, and ensemble) and ten topics emerged based on article abstracts. Underlying both approaches is the consistent recognition of four main areas: text mining/text analytics, machine learning, deep learning, and ensemble methods. This systematic review offers a comprehensive overview of recent text mining research and the emerging trends in this field. It highlights the importance of systematic reviews in synthesizing existing literature and identifying areas for future research.

The aim of the study was to examine the possibility of converting subjective textual data written in the free column space of the Mother and Child Handbook (MCH) into objective information using text mining and to compare any monthly changes in the words written by the mothers. Pregnant women without complications (n = 60) were divided into two groups according to State-Trait Anxiety Inventory grade: low trait anxiety (group I, n = 39) and high trait anxiety (group II, n = 21). Exploratory analysis of the textual data from the MCH was conducted by text mining using the Word Miner software program. Using 1203 structural elements extracted after processing, a comparison of monthly changes in the words used in the mothers' comments was made between the two groups. The data was mainly analyzed by a correspondence analysis. The structural elements in groups I and II were divided into seven and six clusters, respectively, by cluster analysis. Correspondence analysis revealed clear monthly changes in the words used in the mothers' comments as the pregnancy progressed in group I, whereas the association was not clear in group II. The text mining method was useful for exploratory analysis of the textual data obtained from pregnant women, and the monthly change in the words used in the mothers' comments as pregnancy progressed differed according to their degree of unease.

We recently released a plugin to Pathvisio to assist the curator with automatic integration of heterogeneous pathway knowledge. Pathway curation involves gathering this knowledge not only through discussions with peers, but also from biological databases and the scientific literature. Integrating this all into one pathway is challenging and to date still mainly a laborious manual process.More and more biological databases provide webservices (ie. REST or SOAP). The main objective of webservices is to provide an interface for other computer applications. Our plug-in uses these interfaces to gather binary relationships of a given pathway entity and present the relation in star shaped figures to the curator. The plug-in also uses functionality of Pathvisio to project and visualize research data on pathways. If research data exists, this can then be projected on to the suggested potential novel entities. The curator can then make an educated decision on extending pathways with database knowledge.Different text mining methods exist to process large quantities of available literature. Text mining results can be very valuable in pathway curation. Integrating them in a similar way as with biological databases is difficult. It requires a text-mining output standard. In our current plug-in we included text-mining support in our current version of the plug-in.We like to propose here a standard for text mining output. The output standard is a data object that captures the relationships found with text mining, the document identifiers, the sentences from where the results originate, the method used, and its metrics. Such an output standard would enable a user-oriented comparison of text mining results. Currently, text-mining results are mainly evaluated on their performance in finding relevant results. By enabling a text mining out standard, the results become available to the larger user community. The data is immediately accessible to a wide range of applications. If a text mining method supports our output standard, the pathway curation community can use it in Pathvisio through our plug-in.

We recently released a plugin to Pathvisio to assist the curator with automatic integration of heterogeneous pathway knowledge. Pathway curation involves gathering this knowledge not only through discussions with peers, but also from biological databases and the scientific literature. Integrating this all into one pathway is challenging and to date still mainly a laborious manual process.More and more biological databases provide webservices (ie. REST or SOAP). The main objective of webservices is to provide an interface for other computer applications. Our plug-in uses these interfaces to gather binary relationships of a given pathway entity and present the relation in star shaped figures to the curator. The plug-in also uses functionality of Pathvisio to project and visualize research data on pathways. If research data exists, this can then be projected on to the suggested potential novel entities. The curator can then make an educated decision on extending pathways with database knowledge.Different text mining methods exist to process large quantities of available literature. Text mining results can be very valuable in pathway curation. Integrating them in a similar way as with biological databases is difficult. It requires a text-mining output standard. In our current plug-in we included text-mining support in our current version of the plug-in.We like to propose here a standard for text mining output. The output standard is a data object that captures the relationships found with text mining, the document identifiers, the sentences from where the results originate, the method used, and its metrics. Such an output standard would enable a user-oriented comparison of text mining results. Currently, text-mining results are mainly evaluated on their performance in finding relevant results. By enabling a text mining out standard, the results become available to the larger user community. The data is immediately accessible to a wide range of applications. If a text mining method supports our output standard, the pathway curation community can use it in Pathvisio through our plug-in.

This paper develops and estimates an interest rate model with investor attitude factors, which are extracted by a text mining method. First, we consider two contrastive attitudes (optimistic versus conservative) towards uncertainties about Brownian motions driving economy, develop an interest rate model, and obtain an empirical framework of the economy consisting of permanent and transitory factors. Second, we apply the framework to a bond market under extremely low interest rate environment in recent years, and show that our three-factor model with level, steepening and flattening factors based on different investor attitudes is capable of explaining the yield curve in the Japanese government bond (JGB) markets. Third, text mining of a large text base of daily financial news reports enables us to distinguish between steepening and flattening factors, and from these textual data we can identify events and economic conditions that are associated with the steepening and flattening factors. We then estimate the yield curve and three factors with frequencies of relevant word groups chosen from textual data in addition to observed interest rates. Finally, we show that the estimated three factors, extracted only from the bond market data, are able to explain the movement in stock markets, in particular Nikkei 225 index.

Public science is critical to the economy and to society. However, much of the beneficial impact of scientific research only occurs when scientific knowledge is disseminated broadly and is used by others. This thesis examines the emerging policy, law, and practice of facilitating open access to scientific research data. One particular focus is to examine the open data policies recently introduced by research funders and publishers, and the potential in these for driving the practice of open scientific data into the future. This thesis identifies five major stumbling blocks to sustainable open scientific data. Firstly, the prevailing ‘mindset’ that facilitating open access to data is analogous to facilitating open access to publications and, therefore, research data can easily be shared, with research funders and librarians effectively leading the process. Secondly, the unclear meaning of the term ‘data’, which causes confusion among stakeholders. Thirdly, ‘misunderstood incentives’ for data sharing and the additional inputs required from researchers. Fourthly, ‘data privacy’—an issue that only applies to selected research datasets, and yet appears to dominate the discussion about open research data. Finally, there is ‘copyright law’, which poses challenges at different stages of data release and reuse. In this thesis, I argue that the above problems can be addressed using a staged model for open scientific data. I draw specifically on the practice with open scientific data at CERN (the European Organization for Nuclear Research) and the practice of sharing clinical trial data to argue that open data can be shared at various stages of processing and diversification. This model is supplemented by recommendations proposing changes to existing open data mandates and the introduction of a text and data mining exemption into Australian copyright law.

Background: While it is commonly recognized that almost every work and research discipline utilize their own taxonomy, the language used within a specific discipline may also vary depending on numerous factors, including the desired effect of the information being communicated and the intended audience. Different audiences are reached through publication of information, including research results, in different types of publication outlets such as newspapers, newsletters, magazines, websites, and journals. Prior research has shown that students, both undergraduate and graduate, as well as faculty may have a difficult time locating information in different publication outlet types (e.g., magazines, newspapers, journals). The type of publication may affect the ease of understanding and also the confidence placed in the acquired information. A text analytics tool for classifying the source of research as a newsletter (used as a substitute for newspaper articles), a magazine, or an academic journal article has been developed to assist students, faculty, and researchers in identifying the likely source type of information and classifying their own writings with respect to these possible publication outlet types. Literature Review: Literature on information literacy is discussed as this forms the motivation for the reported research. Additionally, prior research on using text mining and text analytics is examined to better understand the methodology employed, including a review of the original Scale of Theoretical and Applied Research system, adapted for the current research. Research Questions: The primary research question is: Can a text mining and text analytics approach accurately determine the most probable publication source type with respect to being from a newsletter, magazine, or journal? Methodology: A text mining and text analytics algorithm, STAR’ (System for Text Analytics-based Ranking), was developed from a previously researched text mining tool, STAR (Scale of Theoretical and Applied Research), that was used to classify the research type of articles between theoretical and applied research. The new text mining method, STAR’, analyzes the language used in manuscripts to determine the type of publication. This method first mines all words from corresponding publication source types to determine a keyword corpus. The corpus is then used in a text analytics process to classify full newsletters, magazine articles, and journal articles with respect to their publication source. All newsletters, magazine articles, and journal articles are from the library and information sciences (LIS) domain. Results: The STAR’ text analytics method was evaluated as a proof of concept on a specific LIS organizational newsletter, as well as articles from a single LIS magazine and a single LIS journal. STAR’ was able to classify the newsletters, magazine articles, and journal articles with 100% accuracy. Random samples from another similar LIS newsletter and a different LIS journal were also evaluated to examine the robustness of the STAR’ method in the initial proof of concept. Following the positive results of the proof of concept, additional journal, magazine, and newsletter articles were used to evaluate the generalizability of STAR’. The second-round results were very positive for differentiating journals and newsletters from other publication types, but revealed potential issues for distinguishing magazine articles from other types of publications. Discussion: STAR’ demonstrates that the language used for transferring information within a specific discipline does differ significantly depending on the intended recipients of the research knowledge. Further work is needed to examine language usage specific to magazine articles. Conclusions: The STAR’ method may be used by students and faculty to identify the likely source of research or discipline-specific information. This may improve trust in the reliability of information due to different levels of rigor applied to different types of publications. Additionally, the STAR’ classifications may be used by students, faculty, or researchers to determine the most appropriate type of outlet and correspondingly the most appropriate type of audience for the reported information in their own manuscripts, thereby improving the chance for successful sharing of information to appropriate audiences who will deem the information to be reliable, through publication in the most relevant outlet type.

Technology roadmapping is a well-established method used in strategy development to map alternative future paths, while text mining offers untapped potentials concerning early detection and environmental scanning. In this paper, the roadmapping process is split into different steps in order to analyse which text mining methods could add further value within each. This leads to a two-layered process model, which includes text mining techniques to systematically integrate external information in ongoing roadmapping processes. Textual data can be used for a structured analysis and exploration of thematic fields and an objective, quantitative summary of actual developments. To demonstrate some of the benefits, the field of "cloud computing" is used to illustrate the procedure. As this article will show, the results provided by this approach extend the existing methodology, integrates an external view and complements expert opinion.

MotivationIn recent years, there has been great progress in the field of automated curation of biomedical networks and models, aided by text mining methods that provide evidence from literature. Such methods must not only extract snippets of text that relate to model interactions, but also be able to contextualize the evidence and provide additional confidence scores for the interaction in question. Although various approaches calculating confidence scores have focused primarily on the quality of the extracted information, there has been little work on exploring the textual uncertainty conveyed by the author. Despite textual uncertainty being acknowledged in biomedical text mining as an attribute of text mined interactions (events), it is significantly understudied as a means of providing a confidence measure for interactions in pathways or other biomedical models. In this work, we focus on improving identification of textual uncertainty for events and explore how it can be used as an additional measure of confidence for biomedical models.ResultsWe present a novel method for extracting uncertainty from the literature using a hybrid approach that combines rule induction and machine learning. Variations of this hybrid approach are then discussed, alongside their advantages and disadvantages. We use subjective logic theory to combine multiple uncertainty values extracted from different sources for the same interaction. Our approach achieves F-scores of 0.76 and 0.88 based on the BioNLP-ST and Genia-MK corpora, respectively, making considerable improvements over previously published work. Moreover, we evaluate our proposed system on pathways related to two different areas, namely leukemia and melanoma cancer research.Availability and implementationThe leukemia pathway model used is available in Pathway Studio while the Ras model is available via PathwayCommons. Online demonstration of the uncertainty extraction system is available for research purposes at http://argo.nactem.ac.uk/test. The related code is available on https://github.com/c-zrv/uncertainty_components.git. Details on the above are available in the Supplementary Material.Supplementary informationSupplementary data are available at Bioinformatics online.

With the overwhelming increase in the amount of texts on the web, it is almost impossible for people to keep abreast of up-to-date information. Text mining is a process by which interesting information is derived from text through the discovery of patterns and trends. Text mining algorithms are used to guarantee the quality of extracted knowledge. However, the extracted patterns using text or data mining algorithms or methods leads to noisy patterns and inconsistency. Thus, different challenges arise, such as the question of how to understand these patterns, whether the model that has been used is suitable, and if all the patterns that have been extracted are relevant. Furthermore, the research raises the question of how to give a correct weight to the extracted knowledge. To address these issues, this paper presents a text post-processing method, which uses a pattern co-occurrence matrix to find the relation between extracted patterns in order to reduce noisy patterns. The main objective of this paper is not only reducing the number of closed sequential patterns, but also improving the performance of pattern mining as well. The experimental results on Reuters Corpus Volume 1 data collection and TREC filtering topics show that the proposed method is promising.

There is global attention on new data analytic methods. Data scraping (a typical first step for advanced data analytics), text and data mining (TDM, the extraction of knowledge from data) and machine learning (ML, often also simply referred to as Artificial Intelligence or AI) are seen as critical technologies. The legal issues involved in the regulation of data range from privacy and data protection (such as the GDPR) to proprietary approaches (such as copyright, database rights, or proposed new rights in data themselves). This paper focusses on one specific intervention, the introduction of two exceptions for text and data mining in the Directive on Copyright in the Digital Single Market (CDSM). Art. 3 is a mandatory exception for text and data mining (TDM) for the purposes of scientific research; Art. 4 permits text and data mining by anyone but with rightsholders able to “contract-out” (Art. 4), for example preventing TDM use of publicly available online content by “machine-readable means”. We trace the context of using the lever of copyright law to enable emerging technologies and support innovation. Within the EU copyright intervention, elements that may underpin a transparent legal framework for AI are identified, such as the possibility of retention of (permanent) copies for further verification. On the other hand, we identify several pitfalls, including an excessively broad definition of TDM which makes the entire field of data-driven AI development dependent on an exception. We analyse the implications of limiting the scope of the exceptions to the right of reproduction (which leaves the communication of research results in a grey zone). We also argue that the limitation of Art. 3 to certain beneficiaries remains problematic; and that the requirement of lawful access is difficult to operationalize. In conclusion, we argue that there should be no need for a TDM exception for the act of extracting informational value from protected works. The EU’s CDSM provisions paradoxically may favour the development of biased AI systems due to price and accessibility conditions for accessing training data that offer the wrong incentives. We also identify some old and new areas of the EU acquis which will play a crucial role in the future relationship of EU copyright law with technological innovation.

Building information modeling (BIM) has emerged as one of the key trends in the architecture, engineering, and construction (AEC) industry. As interest in BIM rapidly increased, the quantity of related literature also increased, thus, it has become important to analyze and identify the key topics and trends in BIM research over this period. Therefore, in order to analyze the research and technology trends related to BIM in Korea, we used the text-mining technique. In addition, by analyzing BIM-related papers using text-mining technology, we can analyze the patterns, main research trends, and trends in specific fields through data preprocessing, which formalizes the unstructured data of sentences, thus presenting us with a specific strategic plan for the future direction of research and technology related to BIM in Korea. In order to propose a strategic direction for future BIM-related research and technology in Korea, in this study, many researches related to BIM in Korea were collected and changes in the patterns and research trends of the BIM research periods were analyzed by dividing them it into the periods of introduction, development, and adaptation using text-mining methods and techniques, frequency analysis, and topic modeling methods. Therefore, in the future, BIM will be developed not only in the field of architecture but in all fields related to the construction industries, such as civil engineering, aviation, and shipping, and more active research will be conducted in various fields.

The objective of this case study is to illustrate how text mining of open-ended responses and sentimental expressions (positive or negative) from student survey could yield valuable information for improving student experience management (SEM). The concept of student SEM was borrowed from the notion of customer experience management (CEM), which aims for ongoing improvement of customer relations through understanding of the customer's point of view. With the advance of text mining technology, which is based upon artificial intelligence and machine learning, textual data that were previously underutilized are found to be valuable in CEM. To illustrate how text mining can be applied to SEM, the authors discuss an example from a campus-wide survey conducted at Arizona State University. The purpose of this survey was to better understand student experiences with instructional technology in order for administrators to make data-driven decisions on its implementation. Rather than imposing the researchers' preconceived suppositions on the students by using force-option survey items, researchers on this project chose to use open-ended questions in order to elicit a free emergence of themes from the students. The most valuable lesson learned from this study is that students perceive an ideal environment as a web of mutually supporting systems. Specifically, online access should be augmented by use of laptops and availability of course materials, whereas virtual classes should be balanced by human interactions.

Recent advances in text mining have provided new methods for capitalizing on the voluminous natural language text data created by organizations, their employees, and their customers. Although often overlooked, decisions made during text preprocessing affect whether the content and/or style of language are captured, the statistical power of subsequent analyses, and the validity of insights derived from text mining. Past methodological articles have described the general process of obtaining and analyzing text data, but recommendations for preprocessing text data were inconsistent. Furthermore, primary studies use and report different preprocessing techniques. To address this, we conduct two complementary reviews of computational linguistics and organizational text mining research to provide empirically grounded text preprocessing decision-making recommendations that account for the type of text mining conducted (i.e., open or closed vocabulary), the research question under investigation, and the data set’s characteristics (i.e., corpus size and average document length). Notably, deviations from these recommendations will be appropriate and, at times, necessary due to the unique characteristics of one’s text data. We also provide recommendations for reporting text mining to promote transparency and reproducibility.

This study proposes a new model for service quality measurement using sentiment analysis and text mining techniques. This model aims to overcome traditional methods' time, cost and implementation difficulties and provide a more dynamic and efficient approach to service quality measurement. In addition, in this model, instead of the dimensions used in service quality measurements, such as SERVQUAL or SERVPERF, it is shown how to determine new categories and keywords specific to the service sector in which the model is used by text mining. Thus, it is aimed at something other than reaching more accurate results in service quality measurement. To achieve the model’s purpose, it aims to develop a service quality measurement model using social media data processed by text mining and sentiment analysis. To find an answer to this question, the keywords "flood", "meter", "rain", "irrigation", "infrastructure", "sewerage", "sewage", "maintenance hole ", "aski", "waterless", "water" were extracted from 109.844 tweets sent to the Twitter account of a municipality between 2016 and 2022 by text mining method. Service quality was measured by subjecting 5766 tweets containing the keywords extracted to sentiment analysis. As a result of the service quality measurement, 1922 negative, 973 positive and 2871 neutral tweets were identified. The average negative score was 0.51, the average positive score was 0.11, and the average neutral score was 0.38.