Search

Cycling is an eco-friendly and sustainable mode of transportation. Despite its benefits, the cyclists’ risk of collision is still high when interacting with other road users. This study analyzed self-reported near-miss and collision event descriptions for the United States provided by the crowdsourcing platform, BikeMaps.org. Innovative and efficient analytic methods are needed to generate useful information from unstructured textual data sources in the transportation domain. In this study, explorative text mining, topic modeling, and machine learning are utilized to gain insights from the unstructured textual descriptions of crowdsourced near-miss and collision events. The approaches are used to unveil prevalent words and word associations for near-miss and collision events. Structural Topic Modeling (STM) is deployed to autogenerate latent themes or topics from the event descriptions. The generated topic proportions are used as input in Artificial Neural Networks (ANN) to estimate the cyclist’s propensity to a collision. It was found that cyclists had a higher propensity to a collision in topics that articulated vehicle encroachment to the bike lane, on-street parking close or into the bike lane resulting in dooring, and drivers’ violations at the crosswalk. The results and methodology used in this study can assist engineers, policymakers, and law enforcement officers to proactively reduce potential cyclist collisions, prioritizing areas where cyclist safety improvements are needed, and ultimately promoting bicycle ridership in our communities.

One of the most significant differences of M5 over previous forecasting competitions is that it was held on Kaggle, an online platform for data scientists and machine learning practitioners. Kaggle provides a gathering place, or virtual community, for web users who are interested in the M5 competition. Users can share code, models, features, and loss functions through online notebooks and discussion forums. Here, we study the social influence of this virtual community on user behavior in the M5 competition. We first research the content of the M5 virtual community by topic modeling and trend analysis. Further, we perform social media analysis to identify the potential relationship network of the virtual community. We study the roles and characteristics of some key participants who promoted the diffusion of information within the M5 virtual community. Overall, this study provides in-depth insights into the mechanism of the virtual community’s influence on the participants and has potential implications for future online competitions.

Topic modeling is a significant branch of natural language processing and machine learning focused on inferring the generative process of text. Traditionally, algorithms for estimating topic models have relied on Bayesian inference and Gibbs sampling. This paper proposes a novel “acceptable set” framework for formulating topic modeling problems inspired by ideas from discrete component analysis and data-driven robust optimization. Our approach not only simplifies the design and inference of topic models, but also allows for extensions and generalizations that are challenging to integrate into traditional approaches. Different restrictions (e.g., sparsity) and assumptions (e.g., alternative generative processes) can be easily incorporated into our formulations through additional or modified constraints. Our formulations also naturally control a widely used metric of model quality, perplexity. We adapt state-of-the-art stochastic gradient methods to find good local optima for the optimization formulations. The algorithms are efficient, scaling to realistic problem sizes with runtimes comparable to existing methods. Through extensive computational experiments, we show that our methods have improved solution quality compared to baseline methods and reconstruct more reliably the underlying generative models. Our framework overcomes known vulnerabilities of traditional topic modeling algorithms: our methods are effective in low-data settings, register good out-of-sample performance, and perform well for a variety of initial assumptions on input parameter values.

Classroom observation is an essential component to improve the quality of teaching and develop educational research. Nonetheless, traditional observation procedures involve previously trained observers, rendering them expensive and time-consuming. Thus, there is a need for a tool which enables to analyze multiple lessons in short time, represent those lessons consistently, and describe them according to the teaching strategies used over time to give teachers timely and continuous feedback. In this work, we propose two unsupervised approaches to represent the evolution of lessons based on topic modeling: textbook- and BERT-based. This way, we can improve the generalization capability of these tools and avoid the need of previously annotated data. Subsequently, we collapsed the derived topics into two super-topics to ease interpretability, and clustered the time series they define to characterize them according to the teaching strategies used. We applied our methodology to 195 science lessons revealing three patterns related to different strategies. Lastly, we compare the two approaches using different metrics for clustering assessment and discuss their benefits and detriments. We expect these models can become useful tools for teachers to obtain an automated and objective representation of their lessons, but also to other stakeholders who want to analyze teaching practices at a large scale.

Deep learning models have achieved high performance across different domains, such as medical decision-making, autonomous vehicles, decision support systems, among many others. However, despite this success, the inner mechanisms of these models are opaque because their internal representations are too complex for a human to understand. This opacity makes it hard to understand the how or the why of the predictions of deep learning models. There has been a growing interest in model-agnostic methods that make deep learning models more transparent and explainable to humans. Some researchers recently argued that for a machine to achieve human-level explainability, this machine needs to provide human causally understandable explanations, also known as causability. A specific class of algorithms that have the potential to provide causability are counterfactuals. This paper presents an in-depth systematic review of the diverse existing literature on counterfactuals and causability for explainable artificial intelligence (AI). We performed a Latent Dirichlet topic modeling analysis (LDA) under a Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework to find the most relevant literature articles. This analysis yielded a novel taxonomy that considers the grounding theories of the surveyed algorithms, together with their underlying properties and applications to real-world data. Our research suggests that current model-agnostic counterfactual algorithms for explainable AI are not grounded on a causal theoretical formalism and, consequently, cannot promote causability to a human decision-maker. Furthermore, our findings suggest that the explanations derived from popular algorithms in the literature provide spurious correlations rather than cause/effects relationships, leading to sub-optimal, erroneous, or even biased explanations. Thus, this paper also advances the literature with new directions and challenges on promoting causability in model-agnostic approaches for explainable AI.

Document clustering allows a user to group semantically similar documents. It has been an interesting research area for the past many years and various methods and techniques have been developed. However, the research has primarily been limited to English and other high resource languages. For low-resource languages, such as Urdu, the area of document clustering is open to contributions. This work presents an experimental evaluation of clustering techniques on Urdu tweets. It is a challenging task to semantically cluster tweets due to their very short length. In this paper, various features, including sentence and phrase-level embeddings, TF-IDF features and document embeddings are extracted from tweets and clustering is performed using three different algorithms: K-Means, Bisecting K-Means, and Affinity Propagation algorithms. Furthermore, a comparison is performed with the traditional topic modeling approach. The results indicate that the TF-IDF features combined with the K-means clustering algorithm outperformed the adopted clustering techniques.

The parallel Hierarchical Dirichlet Process (pHDP) is an efficient topic model which explores the equivalence of the generation process between Hierarchical Dirichlet Process (HDP) and Gamma-Gamma-Poisson Process (G2PP), in order to achieve parallelism at the topic level. Unfortunately, pHDP loses the non-parametric feature of HDP, i.e., the number of topics in pHDP is predetermined and fixed. Furthermore, under the bootstrap structure of pHDP, the topic-indiscriminate words are of high probabilities to be assigned to different topics, resulting in poor qualities of the extracted topics. To achieve parallelism without sacrificing the non-parametric feature of HDP, in addition to improve the quality of extracted topics, we propose a parallel dynamic topic model by developing an adjustment mechanism of evolving topics and reducing the sampling probabilities of topic-indiscriminate words. Both supervised and unsupervised experiments on benchmark datasets show the competitive performance of our model.

The amount of digital text-based consumer review data has increased dramatically and there exist many machine learning approaches for automated text-based sentiment analysis. Marketing researchers have employed various methods for analyzing text reviews but lack a comprehensive comparison of their performance to guide method selection in future applications. We focus on the fundamental relationship between a consumer’s overall empirical evaluation, and the text-based explanation of their evaluation. We study the empirical tradeoff between predictive and diagnostic abilities, in applying various methods to estimate this fundamental relationship. We incorporate methods previously employed in the marketing literature, and methods that are so far less common in the marketing literature. For generalizability, we analyze 25,119 products in nine product categories, and 260,489 reviews across five review platforms. We find that neural network-based machine learning methods, in particular pre-trained versions, offer the most accurate predictions, while topic models such as Latent Dirichlet Allocation offer deeper diagnostics. However, neural network models are not suited for diagnostic purposes and topic models are ill equipped for making predictions. Consequently, future selection of methods to process text reviews is likely to be based on analysts’ goals of prediction versus diagnostics.

Sentiment analysis systems can collect and automatically structure unstructured information by collecting public views of services, products, policy, brands, etc. This information is of major value in the fields of marketing analysis, public relations, product reviews, net promoter evaluations, customer feedback and client reviews. Literary works, on the other hand, are less susceptible to computational analysis because there are no immediate commercial incentives. However, similar techniques can be used to evaluate literary work, comprehend the underlying social network, and obtain or validate literary work. This project is about analyzing the book's characters and predicting their characteristics and relationships with one another. A lot of human effort is expended during the adaptation of a novel/book in any form, which is inconvenient and undesirable. Furthermore, the human brain has a tendency to overlook a number of minor details about the events/characters in the book. The scenario described above can frequently result in inaccuracies in the adaptation's plot. As a result, the project is an innovation that aims to aid in the easy and accurate adaptation of a book, making the process much simpler and precise. Machine learning (supervised and unattended) and lexical approaches include the current sentiment analysis techniques. The model's goal is to scan the massive amounts of text in the book. Following digitization, the model will display interesting ideas derived from the given book using a combination of natural language processing, feelings and emotions analysis, and social network analysis methodology.

Generating explanations within a local and model-agnostic explanation scenario for text classification is often accompanied by a local approximation task. In order to create a local neighborhood for a document, whose classification shall be explained, sampling techniques are used that most often treat the according features at least semantically independent from each other. Hence, contextual as well as semantic information is lost and therefore cannot be used to update a human’s mental model within the according explanation task. In case of dependent features, such explanation techniques are prone to extrapolation to feature areas with low data density, therefore causing misleading interpretations. Additionally, the ”the whole is greater than the sum of its parts” phenomenon is disregarded when using explanations that treat the according words independently from each other. In this paper, an architecture named CaSE is proposed that either uses Semantic Feature Arrangements or Semantic Interrogations to overcome these drawbacks. Combined with a modified version of Local interpretable model-agnostic explanations (LIME), a state of the art local explanation framework, it is capable of generating meaningful and coherent explanations. The approach utilizes contextual and semantic knowledge from unsupervised topic models in order to enable realistic and semantic sampling and based on that generate understandable explanations for any text classifier. The key concepts of CaSE that are deemed essential for providing humans with high quality explanations are derived from findings of psychology. In a nutshell, CaSE shall enable Semantic Alignment between humans and machines and thus further improve the basis for Interactive Machine Learning. An extensive experimental validation of CaSE is conducted, showing its effectiveness by generating reliable and meaningful explanations whose elements are made of contextually coherent words and therefore are suitable to update human mental models in an appropriate way. In the course of a quantitative analysis, the proposed architecture is evaluated w.r.t. a consistency property and to Local Fidelity of the resulting explanation models. According to that, CaSE generates more realistic explanation models leading to higher Local Fidelity compared to LIME.

The agricultural futures prices are generally considered difficult to forecast because the causes of fluctuations are incredibly complicated. We propose a text-based forecasting framework, which can effectively identify and quantify factors affecting agricultural futures based on massive online news headlines. A comprehensive list of influential factors can be formed using a text mining method called topic modeling. A new sentiment-analysis-based way is designed to quantify the factors such as the weather and policies that are important yet difficult to quantify. The proposed framework is empirically tested at forecasting soybean futures prices in the Chinese market. Testing was based on 9715 online news headlines from July 19, 2012 to July 9, 2018. The results show that the identified influential factors and sentiment-based variables are effective, and the proposed framework performs significantly better in medium-term and long-term forecasting than the benchmark model.

Learning hidden topics from data streams has become absolutely necessary but posed challenging problems such as concept drift as well as short and noisy data. Using prior knowledge to enrich a topic model is one of potential solutions to cope with these challenges. Prior knowledge that is derived from human knowledge (e.g. Wordnet) or a pre-trained model (e.g. Word2vec) is very valuable and useful to help topic models work better. However, in a streaming environment where data arrives continually and infinitely, existing studies are limited to exploiting these resources effectively. Especially, a knowledge graph, that contains meaningful word relations, is ignored. In this paper, to aim at exploiting a knowledge graph effectively, we propose a novel graph convolutional topic model (GCTM) which integrates graph convolutional networks (GCN) into a topic model and a learning method which learns the networks and the topic model simultaneously for data streams. In each minibatch, our method not only can exploit an external knowledge graph but also can balance the external and old knowledge to perform well on new data. We conduct extensive experiments to evaluate our method with both a human knowledge graph (Wordnet) and a graph built from pre-trained word embeddings (Word2vec). The experimental results show that our method achieves significantly better performances than state-of-the-art baselines in terms of probabilistic predictive measure and topic coherence. In particular, our method can work well when dealing with short texts as well as concept drift. The implementation of GCTM is available at \url{https://github.com/bachtranxuan/GCTM.git}.

Aiming at disclosing a general research landscape of structural engineering in the twenty-first century, this study applies the latent Dirichlet allocation (LDA), a topic modeling approach, to analyze 51,346 article abstracts from 23 prestigious journals in structural engineering with a publication period from 2000 to 2020. The LDA analyzes the literature inventory by extracting 50 distinguishable wordclouds, each centered around one distinct research theme and assigned a unique topic name. Subsequently, various measures have been proposed to integrate the posterior distributions of these research topics with article information such as publication year, journal name, and correspondence address. The increase index identifies five cold and hot topics, which reflect the shift of research interests in the community. Emerging research topics such as seismic risk assessment and composite material have received much more attention in recent years. Moreover, advanced metrics have been proposed to analyze the research similarity and evolution across different journals and countries/regions. As discussed in the paper, analysis findings would enable community stakeholders (e.g., students, engineers, researchers, conference organizers, journal editors, funding agencies) to explore the state of the research and develop viable strategies to further foster the healthy growth of the community. Such strategies can be (1) researchers submitting a paper to the most appropriate journal; (2) journal editors adjusting the journal focus to enhance its impact; and (3) funding agencies prioritizing research supports that best fit regional needs and circumstances, among others.

Blood pressure (BP) is an important indicator of an individual's health status and is closely related to daily behaviors. Thus, a continuous daily measurement of BP is critical for hypertension control. To assist continuous measurement, BP prediction based on non-physiological data (ubiquitous mobile phone data) was studied in the research. An algorithm was proposed that predicts BP based on patients’ daily routine, which includes activities such as sleep, work, and commuting. The aim of the research is to provide insight into the application of mobile data in telemonitoring and the continuous unobtrusive daily measurement of BP. A half-year data set from October 2017 of 320 individuals, including telecom data and BP measurement data, was analyzed. Two hierarchical Bayesian topic models were used to extract individuals’ location-driven daily routine patterns (topics) and calculate probabilities among these topics from their day-level mobile trajectories. Based on the topic probability distribution and patients’ contextual data, their BP were predicted using different models. The prediction model comparison shows that the long short-term memory (LSTM) method exceeds others when the data has a high dependency. Otherwise, the Random Forest regression model outperforms the LSTM method. Also, the experimental results validate the effectiveness of the topics in BP prediction.

Sparse and short news headlines can be arbitrary, noisy, and ambiguous, making it difficult for classic topic model LDA (latent Dirichlet allocation) designed for accommodating long text to discover knowledge from them. Nonetheless, some of the existing research about text-based crude oil forecasting employs LDA to explore topics from news headlines, resulting in a mismatch between the short text and the topic model and further affecting the forecasting performance. Exploiting advanced and appropriate methods to construct high-quality features from news headlines becomes crucial in crude oil forecasting. This paper introduces two novel indicators of topic and sentiment for the short and sparse text data to tackle this issue. Empirical experiments show that AdaBoost.RT with our proposed text indicators, with a more comprehensive view and characterization of the short and sparse text data, outperforms the other benchmarks. Another significant merit is that our method also yields good forecasting performance when applied to other futures commodities.

International audience; We introduce a deep latent recommender system named deepLTRS in order to provide users with high quality recommendations based on observed user ratings \textit{and} texts of product reviews. The underlying motivation is that, when a user scores only a few products, the texts used in the reviews represent a significant source of information, thereby enhancing the predictive ability of the model. Our approach adopts a variational auto-encoder (VAE) architecture as a deep generative latent model for an ordinal matrix encoding ratings and a document-term matrix encoding the reviews. Taking into account both matrices as model inputs, deepLTRS uses a neural network to capture the relationship between latent factors and latent topics. Moreover, a user-majoring encoder and a product-majoring encoder are constructed to jointly capture user and product preferences. Due to the specificity of the model structure, an original row-column alternated mini-batch optimization algorithm is proposed to deal with user-product dependencies and computational burden. Numerical experiments on simulated and real-world data sets demonstrate that deepLTRS outperforms the state-of-the-art, in particular in context of extreme data sparsity.

With the massive use of electronic gadgets and the developing fame of web-based media, a great deal of text information is being produced at the rate never observed. It is not feasible for people to pursue all information produced and discover what is being investigated in their area of interest. To determine topics in large textual documents Topic modelling is used. Topic Modeling Algorithms (TMA) are Unsupervised Machine Learning approaches which are widely used and have proven to be successful in the area of Aspect-based Opinion Mining to extract ‘latent’ topics, which are aspects of interest. In this paper, the approaches that are widely used for topic modelling are examined and compared to find their importance in detecting topics based on metrics such as perplexity and coherence. As a result, Latent Dirichlet Allocation is a good topic modelling algorithm compared to Latent Semantic Analysis and Hierarchical Dirichlet Process for aspect extraction process in aspect-based opinion mining. Also, we have proposed an unsupervised aspect extraction algorithm based on topic models for Aspect-based Opinion mining.

It is necessary to reduce the cognitive overhead of interpreting the native topic term list of the Latent Dirichlet Allocation (LDA) style topic model. In this regard, automatic topic labeling has become an effective approach to generate meaningful alternative representations of topics discovered for end-users. In this study, we introduced a novel two-phase neural embedding framework with the redundancy-aware graph-based ranking process. It demonstrated how pre-trained neural embedding could be usefully applied in topic terms, sentence presentations, and automatic topic labeling tasks. Moreover, reranking the topic terms optimized the discovered topics with fewer yet more representative terms while retaining the topic information integrality and fidelity. It further decreased the burden of computation caused by neural embedding and improved the overall effectiveness of the labeling system. Compared with the prevailing state-of-the-art and classical labeling systems, our efficient model boosted the quality of the topic labels generated and discovered more meaningful topic labels.

Learning from past incidents is critical to achieving and maintaining high process safety performance. Accident and incident records provide one way for learning; however, these are usually in the form of unstructured texts, which makes analysis difficult. Recently, text mining methods based on supervised learning have been proposed for analyzing accident reports; however, they require an impractically large number of labeled records as training examples. This paper proposes an automated, semi-supervised, domain-independent approach for analyzing accident reports. Given a set of user-defined classification topics and domain literature such as handbooks, glossaries, and Wikipedia articles, the method can identify domain-specific keywords and group them into topics with minimal expert involvement. These keywords and topics can then be used for various data mining purposes, including classification. The proposed approach is demonstrated using two different case studies across domains: (1) in aviation to identify the stage of flight when an accident occurs, and (2) in the process industry domain to identify the cause of pipeline accidents. The average classification accuracy of the proposed method was 80% which is comparable to that of supervised learning methods. The key benefits of this approach are that it can generate domain-specific predictive models with limited manual intervention.

Recently, listwise collaborative filtering (CF) algorithms are attracting increasing interest due to their efficiency and prediction quality. Different from rating-oriented (pointwise) CF, they recommend a preference ranking of items to each user without estimating the absolute value of the ratings. In practice, there are extensive side information about users and items in many corpora that can improve recommendation performance and specially help CF methods to address the cold-start problem. However, a model has not been proposed that incorporates side information to listwise CF. Therefore, in this work, a Bayesian graphical model, called Content-Aware Listwise Collaborative Filtering (CALCF), is developed which incorporates text information, represented as a bag of words, to listwise CF using topic models. We propose a Gibbs sampler with closed-form samples using data augmentation techniques to infer the latent variables. CALCF has been validated by comparison with previous listwise and pointwise algorithms on implicit (click, view, purchase) and explicit (numeric ratings) feedback data in both cold-start and warm-start scenarios. The results demonstrate that in most cases CALCF achieves better Normalized Discounted Cumulative Gain (NDCG) and Recall at top M recommendation compared with the previous models.

As an important strategic emerging industry to promote sustainable economic development, new energy vehicles (NEVs) play multiple vital roles in energy saving and emission reduction and promote the transformation and upgrading of the automobile industry. China's government has formulated abundant policies to support the NEV industry. Existing studies do not pay enough attention to the latent information from NEV policy documents. This paper explores the latent topics of numerous NEV policy documents and their impact on promoting NEVs at the city level by combining the Latent Dirichlet Allocation (LDA) topic model and the econometric method. The results show that the latent topics of NEV policy documents can be categorized into charging infrastructure operation, promotion subsidy, and production support. The prevalence of the promotion subsidy topic is the strongest among the three types of topics. Total topic prevalences significantly promote NEV sales. The 2013 year is a turning point in policy preferences, and this change boosts NEV sales. The prevalence of promotion subsidy topic has an inverted U-shaped effect on NEV sales. The NEV sales will increase by 19% when the prevalence of production support topic increases by 1. Under the COVID-19, although the promotion subsidy policy is still dominant, the production support policy has been paid more attention. Finally, combining the development of China's NEV industry under the COVID-19 epidemic, the paper puts forward pertinent policy suggestions.

Many B2B firms have widely accepted AI-based chatbots to provide human-like service interaction at different customer touchpoints in recent years. One of the objectives behind introducing this technology is to provide an enhanced, live channel Customer Experience (CX) all round the clock. Researchers have focused on delivering the CX by improvising the chatbot's internal algorithm, giving limited attention to CX theories from management literature, which leaves a gap. With the proposed paper, we have investigated the influencing factors of AI-based chatbots from the lens of CX theories for B2B firms. In this paper, a model for organizing CX has been proposed using the diffusion of innovation theory, trust commitment theory, information systems success model, and Hoffman & Novak's flow model for the computer-mediated environment and verified using the social media data. The methodology used for this study is the social media analytics-based content analysis method (sentiment analysis, hierarchical clustering, topic modeling) for data preparation, followed by lasso and ridge regression for model verification. The results suggest that CX in B2B enterprises using chatbots is influenced by these bots' overall system design, customers' ability to use technology, and customer trust towards brand and system.

The goal of this article is twofold. On the one hand I want to outline the main topics, themes and trends of the past 123 (at the time of writing) issues of Russian Literature and shed some light on the journal’s main subject areas. On the other hand, I show how ‘traditional’ topic modeling visualizations such as topic distributions per document, word distributions per topic, and time series of topic distributions, can be complemented by novel approaches such as the topic map, which significantly facilitates the interpretation of the corpus structure and document similarities. Moreover, topic maps reduce the need to make sense of individual topics and allow the researcher to focus directly on the texts themselves and on the corpus as a whole.

The investigation of user preferences through user comments has attracted significant attention. Although topic models have been verified as useful tools to facilitate the understanding of textual contents, they cannot be directly applied to accomplish this task because of two problems. The first problem is the severe data sparsity suffered by user comments because they are generally short. The second problem is the mixture of opinions from both user comments and the original documents the users commented on. To simultaneously solve the data sparsity problem and explore clean user preferences, we propose an author co-occurring topic model (AOTM) for normal documents and their short user comments. By considering authorship, AOTM allows each author of short texts to have a probability distribution over a set of topics represented only short texts. Accordingly, the individual user preferences can be investigated based on these author-level distributions. We verify the performance of AOTM using two news article datasets and one e-commerce dataset. Extensive experiments demonstrate that the AOTM outperforms several state-of-the-art methods in topic learning and topic representation of documents. The potential usage of AOTM in exploring individual user preferences is further illustrated by drawing user portraits and predicting user posting behaviors.

Objective: We report a survey in regional Queensland to understand the reasons for suboptimal uptake of the COVIDSafe app. Methods: A short five‐minute electronic survey disseminated to healthcare professionals, mining groups and school communities in the Central Queensland region. Free text responses and their topics were modelled using natural language processing and a latent Dirichlet model. Results: We received a total of 723 responses; of these, 69% had downloaded the app and 31% had not. The respondents’ reasons for not downloading the app were grouped under four topics: lack of perceived risk of COVID‐19/lack of perceived need and privacy issues; phone‐related issues; tracking and misuse of data; and trust, security and credibility. Among the 472 people who downloaded the app and provided text amenable to text mining, the two topics most commonly listed were: to assist with contact tracing; and to return to normal. Conclusions: This survey of a regional population found that lack of perceived need, concerns around privacy and technical difficulties were the major barriers to users downloading the application. Implications for public health: Health promotion campaigns aimed at increasing the uptake of the COVIDSafe app should focus on promoting how the app will assist with contact tracing to help return to ‘normal’. Additionally, health promotors should address the app's impacts on privacy, people's lack of perceived need for the app and technical barriers.

Organising in the context of projects has been extensively investigated with a focus on project-level processes. However, the industry and market impact, i.e. the field influencing potential of this temporary form of organizing, is still poorly understood. By using concepts derived from the literature on field-configuring events (FCE), we analyse the influence of two projects on the field of electric mobility in Germany. To capture the role of projects within fields, we apply a longitudinal mixed-methods approach combining interpretative qualitative methods with a topic modeling machine learning research strategy. The results highlight how projects can shape fields by opening and closing public debates and narratives in the sense of discursive institutional work. Additionally, we offer concepts that enable a more complete understanding of the sequential entanglement between projects and events, based on three scenarios: the project-by-itself, project-follows-event, and event-follows-project scenarios. With these insights, our empirical study contributes to the project management and institutional literature.

Autonomous vehicle (AV) technologies have been rapidly advancing. One benefit of AVs is that the technology could eliminate many driver errors and also mitigate many pedestrian and bicyclist collisions. Real-world AVs have been tested in many cities. Five companies are running around 50 AVs in Pittsburgh, following the autonomous testing guidelines. BikePGH, a non-profit organization located in Pittsburgh, Pennsylvania conducted a follow-up survey in 2019 (the first survey was conducted in 2017) to understand non-motorists’ opinions of AVs. This study examined how pedestrians and bicyclists perceived AV safety based on their understanding and experiences. At first, this study performed a comparison group test to determine which questions vary by participants’ AV safety rating. The responses were later analyzed with a data mining method known as ‘association rules mining.’ A new performance measure, known as the rule power factor, was then used to identify the significant patterns in the form of rules. The participants also provided their thoughts in responses to the open-ended questions. Using Latent Dirichlet Allocation (LDA), a topic modeling algorithm, 40 topic models were developed based on five open-ended questions. The findings show that the non-motorists showed comparatively fewer negative opinions towards AVs than positive assessments. The results also show that perception patterns vary by the participant’s rating on AV safety. Findings of this study would be beneficial for the AV stakeholders in making AVs and roadways safer for non-motorists.

An innovative model-based approach to coupling text clustering and topic modeling is introduced, in which the two tasks take advantage of each other. Specifically, the integration is enabled by a new generative model of text corpora. This explains topics, clusters and document content via a Bayesian generative process. In this process, documents include word vectors, to capture the (syntactic and semantic) regularities among words. Topics are multivariate Gaussian distributions on word vectors. Clusters are assigned corresponding topic distributions as their semantics. Content generation is ruled by text clusters and topics, which act as interacting latent factors. Documents are at first placed into respective clusters, then the semantics of these clusters is then repeatedly sampled to draw document topics, which are in turn sampled for word-vector generation. Under the proposed model, collapsed Gibbs sampling is derived mathematically and implemented algorithmically with parameter estimation for the simultaneous inference of text clusters and topics. A comparative assessment on real-world benchmark corpora demonstrates the effectiveness of this approach in clustering texts and uncovering their semantics. Intrinsic and extrinsic criteria are adopted to investigate its topic modeling performance, whose results are shown through a case study. Time efficiency and scalability are also studied.

Study objective Topic identification can facilitate knowledge curation, discover thematic relationships, trends, and predict future direction. We aimed to determine through an unsupervised, machine learning approach to topic modeling the most common research themes in emergency medicine over the last 40 years and summarize their trends and characteristics. Methods We retrieved the complete reference entries including article abstracts from Ovid for all original research articles from 1980 to 2019 within emergency medicine for six widely-cited journals. Abstracts were processed through a natural language pipeline and analyzed by a latent Dirichlet allocation topic modeling algorithm for unsupervised topic discovery. Topics were further examined through trend analysis, word associations, co-occurrence metrics, and two-dimensional embeddings. Results We retrieved 47,158 articles during the defined time period that were filtered to 20,528 articles for further analysis. Forty topics covering methodologic and clinical areas were discovered. These topics separated into distinct clusters when embedded in two-dimensional space and exhibited consistent patterns of interaction. We observed the greatest increase in popularity in research themes involving risk factors (0.4% to 5.2%), health utilization (1.2% to 5.0%), and ultrasound (0.7% to 3.3%), and a relative decline in research involving basic science (8.9% to 1.1%), cardiac arrest (6.5% to 2.2%), and vitals (6.3% to 1.3%) over the past 40 years. Our data show only very modest growth in mental health and substance abuse research (1.0% to 1.6%), despite ongoing crises. Conclusions Topic modeling via unsupervised machine learning applied to emergency medicine abstracts discovered coherent topics, trends, and patterns of interaction.

Current research on subset selection for opinion analysis assumes that their methods can retrieve the opinions expressed in documents from general text features. However, such relaxed conditions can hardly maintain the performance of the analysis in opinion mining , especially when given strict limitations on the subset size. In this paper, we propose a framework for opinion subset selection. This framework can select a small set of instances from original data to convey a subjective representation for opinion classification and regression. Compared with our framework, the conventional submodular based subset selection approach cannot capture the fine-grained opinion features expressed in the corpus. Specifically, we propose a monotone non-decreasing score function and a framework based on topic modeling and submodular maximization for filtering irrelevant information and selecting the subsets. Our work further introduces an opinion-sensitive algorithm for optimizing the proposed function for opinion subset construction . We perform extensive experiments and comparative analysis of different subset selection methods in this work. The experimental result shows that the proposed opinion subset selection framework can compress the original text training set and preserve the test set’s classification and regression metric performance at the same time.

This study analyses the agenda setting on social media in the COVID-19 pandemic by exploiting one of the disruptive technologies, big data analytics. Our purpose is to examine whether the agenda of news organisations matches the public agenda on social media in crisis situations, and to explore the feasibility and efficacy of applying big data analytics on social media data. To this end, we used an unsupervised machine learning approach, structural topic modelling and analysed 129,965 tweets posted by UK news media and citizens during April 2, and 8, 2020. Our study reveals a wide diversity of topics in the tweets generated by both groups and finds only a small number of topics are similar, indicating different agendas set in the pandemic. Moreover, we show that citizen tweets focused more on expressing feelings and sharing personal activities while news media tweets talked more about facts and analysis on COVID-19. In addition, our results find that citizens responded more significantly to breaking news. The findings of the study contribute to the agenda setting literature and offer valuable practical implications.

With the fast development of Internet technology, more and more online social networks are changing our daily life. Whether or not an accurate recommendation can be provided for each user in the massive amount of information directly affects the user’s enthusiasm for receiving network services and the user experience effect, which in turn determines the user’s participation and loyalty to network applications. However, most previous methods only use a single network topology information and ignore other auxiliary information (such as user content information). Moreover, how to deal with large scale network is a challenging task. To tackle these challenges, we propose a topic-aware network embedding approach for providing intelligent recommendation services. Specifically, we first extract the network topology based on the constructed social network. Then, we extract the topic information based on the context released by the users with the help of topic model. Finally, a topic-aware network embedding framework is utilized for recommendation. Experimental results on two-widely used dataset demonstrate that our method can achieve the best performance.

In this paper, we propose GMoT-Seq2Seq, a sequence to sequence (Seq2Seq) model with a gated mixture of topics (MoT) designed to utilize topic information to generate fluent and coherent responses. Seq2Seq model is good at capturing the local structure of word sequence which affects the fluency due to their sequential nature, but probably has difficulty to extract topic information from the utterance. In contrast, topic models are very capable of capturing global semantic information that has a direct impact on the coherence. Absorbing the advantages of both, the proposed GMoT-Seq2Seq model uses a Seq2Seq to capture the temporal dependencies, and an MoT layer to obtain the topic vector that provides global semantic dependencies in the conversation. The MoT layer can summarize the utterances into a proportion vector over several underlying topics. To balance the fluency and coherence, we utilize a topic gate to dynamically control the information from the inferred topic vector and the partially generated responses. Experiment results show that our proposed model outperforms the compared baselines, and can generate more fluent and coherent responses.

Summary This paper investigates how ExxonMobil uses rhetoric and framing to shape public discourse on climate change. We present an algorithmic corpus comparison and machine-learning topic model of 180 ExxonMobil climate change communications, including peer-reviewed publications, internal company documents, and advertorials in The New York Times. We also investigate advertorials using inductive frame analysis. We find that the company has publicly overemphasized some terms and topics while avoiding others. Most notably, they have used rhetoric of climate "risk" and consumer energy "demand" to construct a "Fossil Fuel Savior" (FFS) frame that downplays the reality and seriousness of climate change, normalizes fossil fuel lock-in, and individualizes responsibility. These patterns mimic the tobacco industry's documented strategy of shifting responsibility away from corporations—which knowingly sold a deadly product while denying its harms—and onto consumers. This historical parallel foreshadows the fossil fuel industry's use of demand-as-blame arguments to oppose litigation, regulation, and activism.

The servitization domain consists of over three decades of multi-disciplinary research on service activities in industrial contexts. Servitization literature combines different research streams that share a set of critical concepts. Existing meta-analytical studies have organized information content of sub-streams, homogenized theoretical propositions, and concepts to discover shared patterns, and identified an implicit meta-narrative. This study reverses the meta-analysis direction to deconstruct the servitization body of knowledge using the dynamic topic modeling (DTM) methodology to analyze 550 research articles. DTM enables complex forms of content analysis that combine quantitative and qualitative analysis. The analysis demonstrates how these streams have informed the development of the servitization domain and shaped the collective construction of this body of knowledge. The contributions of this study are threefold. First, the study increases understanding of the conceptual dynamics and thematic trendswithin the servitization research domain and the nuances between the sub-streams. The study offers some strategies for the future development of the field, facilitating the renewal of the servitization-related research agenda. Second, it illustrates the role of DTM as an alternative tool for conducting a literature review. Finally, it supports the development of a common language for the servitization field, thereby reducing the entry barriers for new contributors and favoring the knowledge transfer to professionals.

Maritime transport is considered as the lynchpin of global trade. In the past decades, a large number of academic articles in the maritime transport field have been published. This study aims to provide a comprehensive review of the development of maritime transport research. A structural topic model (STM) is adopted to analyze the research themes and trends in the maritime literature. STM is a text mining-based methodology to uncover main topics from large-scale unstructured textual data. In total, 3199 articles published between Jan 1991 and Aug 2020 were collected and analyzed. The results show that the major academic concerns of maritime transport are related to port management, container operations and liner shipping management. The overall research trends have shifted from regulations and policy management to efficient, integrated and sustainable maritime transport over the past few decades. The main research topics and emerging trends discovered from STM can facilitate researchers, funding groups and policymakers for identifying contemporary research issues and making more informed decisions. This research is supported by the National Natural Science Foundation of China (NSFC) (Grant Nos. 72001123, 72031005), by Key project of Philosophy and Social Science Research of Ministry of Education, PRC (Grant No. 20JZD016), and bya grant from Center for Smart Logistics and Supply Chain Systems, Tsinghua University, China (Project No. THUCSL20182911756-010).

The story is the description of events in chronological order that have occurred between people. By delivering facts to the people reading the story, it enables them to feel emotions. Such a story is composed using the following method: each event is analyzed and a storyline is composed, which becomes a skeleton text by linking relationships between major events. As the content of users becomes more diverse, multimodal story composition has become more essential than unimodal text-based story composition. This paper discusses modality integration based on multimodal data types and type conversion for multimodal story composition. We propose a story-graph model to create a story based on the integrated analysis of various modal data. In terms of architecture, the proposed multimodal storytelling model consists of modal data and a topic modeling module that performs clustering based on cross-modal similarities and extracts a topic of clustered modalities. From the perspective of utilization, to visualize a story-graph, the proposed model summarizes nodes with a representative image. Furthermore, the latest techniques are discussed with respect to five main modules and twelve sub-modules for story composition. Lastly, problems that can become issues when composing multimodal story modules are explained.

In recent years, XACML (eXtensible Access Control Markup Language) has been widely used in the development of various applications, especially Web services. The evaluation time of a PDP (Policy Decision Point) grows significantly when the PDP loads a large-scale policy set coded in XACML. In order to improve the PDP evaluation performance, we propose an optimized policy evaluation engine, namely XDLEngine, and make the following contributions. First, XDLEngine has an advantage in the process of handling a large-scale policy set, and innovatively adopts the LDA (Latent Dirichlet Allocation) topic model to cluster policies. Second, according to the clustering results of the LDA model, we digitize and vectorize all rules in policy sets, which facilitates the rule matching. Third, the cosine similarity is introduced to classify the rules under each topic, which greatly reduces the number of comparisons in the process of rule matching and improves the matching efficiency of XDLEngine. Finally, due to the independence between different topics, we use a multi-threaded parallel search in the process of rule matching, which significantly lowers the evaluation time of XDLEngine. The experimental results show that when the number of requests reaches 20,000, the evaluation time of XDLEngine for a practical large-scale policy set with 120,000 rules is approximately 2.48%, 3.47% and 3.68% of that of the Sun PDP, XEngine and HPEngine, respectively.

Background & Aims: Depression is a common and sometimes severe form of mental illness, and public attitudes towards depression can impact the psychological and social functioning of depressed patients. The purpose of the present study was to investigate public attitudes toward depression and three-year trends in these attitudes using big data analysis of social media posts in China. Methods: A search of publically available Sina Weibo posts from January 2014 to July 2017 identified 20,129 hot posts with the keyword term “depression”. We first used a Chinese Linguistic Psychological Text Analysis System (TextMind) to analyze linguistic features of the posts. And, then we used topic models to conduct semantic content analysis to identify specific themes in Weibo users’ attitudes toward depression. Results: Linguistic features analysis showed a significant increase over time in the frequency of terms related to affect, positive emotion, anger, cognition (including the subcategory of insight), and conjunctions. Semantic content analysis identified five common themes: severe effects of depression, stigma, combating stigma, appeals for understanding, and providing support. There was a significant increase over time in references to social (as opposed to professional) support, and a significant decrease over time in references to the severe consequences of depression. Conclusions: Big data analysis of Weibo posts is likely to provide less biased information than other methods about the public's attitudes toward depression. The results suggest that although there is ongoing stigma about depression, there is also an upward trend in mentions of social support for depressed persons. A supervised learning statistical model can be developed in future research to provide an even more precise analysis of specific attitudes.

Topic modeling methods have usually been applied in the past to identify the research interests of researchers. Observing the scientific growth, the trending topics can be identified as Stable, Hot, or Cold. Finding rising stars (junior researchers, who are at the start of their career) from a bibliometric network is a challenging task, specifically if the researchers have an interest in multiple sub-domains or are working on diverse topics. Existing methods for finding rising stars explore the co-author networks or citation networks, and ignore the textual content, which may help in finding rising stars through hot topics detection over time. A publication contributing to a hot topic can be an indication that the author of that publication may be a rising star and can become an expert in that domain in the future. This study proposes the Hot Topics Rising Star Rank (HTRS-Rank) method for finding rising stars by detecting hot topics. HTRS-Rank finds the junior scholars, who contribute to hot topics at the start of their career and ranks them based on the presence of hot topics in their publications. AMiner five years dataset ranging from 2005–2009 is selected for experimentation. Top 10 researchers are considered to measure the association strength using rank correlation among HTRS-Rank and baseline methods. Experimental results show the efficiency of HTRS-Rank in comparison to the baseline methods. The proposed HTRS Rank (TF–IDF) provides low standard deviation for productivity, citations and sociality as compared to baseline methods for more social and highly cited authors. It is identified that HTRS-Rank (WordNet) emphasizes the semantic similarity of two sentences, whereas HTRS-Rank (TF–IDF) scheme emphasizes the uniqueness or importance of each term, therefore TF–IDF approach performs better than WordNet approach due to having higher correlation with StarRank and WMIRank.

Topic modeling is an unsupervised task which helps to capture hidden semantics structure of words in a document. It is significant area of research due to its diverse applications in multidisciplinary fields such as materials sciences, chemical engineering, industrial aspects, social media analysis, bioinformatics and many more. In materials science and engineering, with the help of topic modeling methods, clustering of keywords related to material synthesis steps can be performed which further leads to the detection of topics related to various steps. e.g. keywords related to synthesis procedures can be clustered into topics such as centrifuging, grinding, heating, and dissolving. Classification technique may be used on these topics to make groups of materials based upon their synthesis steps such as hydrothermal, or solid-state. In this paper, an overview and evolution of prominent topic modeling techniques is presented. Also, applications of topic modeling in materials science are discussed.

The Government of DKI Jakarta published transportation integrated program named Jak Lingko to improve public transportation system in Jakarta. This program covers several types of mass transportation in Jakarta such as Mass Rapid Transit (MRT), Light Rail Transit (LRT), Electric Train (KRL Commuter Line), and Bus Transjakarta (TJ). In hope, the people will be interested to switch their modes of transportation to using public transportation considering the congestion and air pollution in Jakarta which is horrible. By utilizing data from twitter, this study tries to analyze public opinion on public transportation by classifying theirs opinions into positive & negative. Furthermore, topic modeling activites will be conducted using the Latent Dirichlet Allocation (LDA) method to find out topics that are often discussed publicly for each public transportation. It was found that the KRL is the only type of public transportation with the number of negative responses exceeding the number of positive responses. It was found that in the MRT & LRT, the community tended to mention words that had positive sentiments, whereas in KRL & TJ people tended to mention words that had negative sentiments. In the grouping of public opinion, the LDA model produced a topic of public discussion on public transportation that was different for each type of public transportation. The model can also capture public communication with the official account administration of public transportation. Prediction models obtained the lowest error with a Mean Average Percentage Error of 8.79%. By adding the public sentiment variable, it can improve the accuracy of the prediction model, but the tweets sentiment variable has no direct relationship to the variable number of public transportation users.

In the digital and Internet era, companies are racing to profile their target users based on their online activities. One of the reliable sources is the news articles they read that can represent their interests. However, extracting latent information from the news articles is not an easy task for a human. In this paper, we introduced a practical model to automatically extract latent information from news articles with pre-determined topics. Our proposed model used unsupervised learning, thus alleviating the need for humans to label news items manually. Doc2vec was used to generate word vectors for each article. Afterward, a spectral clustering algorithm was applied to group the data based on the similarity. A supervised Long Short Term Memory (LSTM) model was built to compare the clustering performance. The best 1, best 3, and best 5 scores were used to evaluate our model. The result showed that our model could not outperformed LSTM model for the best 1 score. However, the best 5 score result indicated that our model was sufficiently robust to cluster the articles based on topic similarity. Additionally, the proposed unsupervised model was implemented in both an on-premise server, and a cloud server. Surprisingly, our proposed method could run faster in the cloud server despite its less number of CPU cores.

Topic modeling is an unsupervised machine learning technique for finding abstract topics in a large collection of documents. It helps in organizing, understanding and summarizing large collections of textual information and discovering the latent topics that vary among documents in a given corpus. Latent Dirichlet allocation (LDA) and Non-Negative Matrix Factorization (NMF) are two of the most popular topic modeling techniques. LDA uses a probabilistic approach whereas NMF uses matrix factorization approach, however, new techniques that are based on BERT for topic modeling do exist. In this paper, we aim to experiment with BERTopic using different Pre-Trained Arabic Language Models as embeddings, and compare its results against LDA and NMF techniques. We used Normalized Pointwise Mutual Information (NPMI) measure to evaluate the results of topic modeling techniques. The overall results generated by BERTopic showed better results compared to NMF and LDA.

Estimating the number of latent topics is a prerequisite to using topic modeling algorithms such as LDA. A number of methods for this purpose are compared, including the de facto approach based on perplexity. Results from synthetic data show that a combination of methods yields better estimates than any single method, and in particular it provides an indicator of the reliability of the estimated number of topics, something no single method can do. In this paper, we extend the findings based on synthetic data over real data and introduce a technique, “Cosine inter-intra topics”, to assess the validity of the ground truth of real data. Moreover, we measure the reliability of estimated number of topics through the variance of methods. Results corroborate the ones from synthetic data, suggesting the approach generalizes to real text corpus.

International audience; The rapid advancement of artificial intelligence (AI) offers exciting opportunities for marketing practice and academic research. In this study, through the application of natural language processing, machine learning, and statistical algorithms, we examine extant literature in terms of its dominant topics, diversity, evolution over time, and dynamics to map the existing knowledge base. Ten salient research themes emerge: (1) understanding consumer sentiments, (2) industrial opportunities of AI, (3) analyzing customer satisfaction, (4) electronic word-of-mouth–based insights, (5) improving market performance, (6) using AI for brand management, (7) measuring and enhancing customer loyalty and trust, (8) AI and novel services, (9) using AI to improve customer relationships, and (10) AI and strategic marketing. The scientometric analyses reveal key concepts, keyword co-occurrences, authorship networks, top research themes, landmark publications, and the evolution of the research field over time. With the insights as a foundation, this article closes with a proposed agenda for further research.

In the era of social media and travel websites, tourists increasingly post reviews about their travel experiences. These posts influence the travel plans and decisions of new tourists and hence; their content is important for businesses and governments. The purpose of this study is to propose a model to investigate tourists’ perceptions and underlying reasons for posting respective content on travel websites. A web analytics-based approach is proposed to conduct in-depth analyses of the data extracted from such websites and generate meaningful insights. A combination of sentiment analysis and topic modeling through the Latent Dirichlet Allocation and machine learning algorithm is used for analyzing the data. The application of the proposed model is illustrated in the case of Goa, India. Findings reveal that the lack of cleanliness, safety, parking, price, facilities, and services are pertinent topics leading to negative sentiments. These topics also corroborate the social and economic events that took place during the period of study. Further clustering of tourist destinations across Goa reveals the presence of similar dissatisfiers. The above findings can be used for deciding the strategy for remedial action by relevant stakeholders. Moreover, the model can automatically classify and analyze future reviews in real-time.

Moro, S., Pires, G., Rita, P., & Cortez, P. (2020). A cross-cultural case study of consumers' communications about a new technological product. Journal of Business Research, 121(December), 438-447. [Advanced online publication on 16 august 2018] Link: https://doi.org/10.1016/j.jbusres.2018.08.009. DOI: 10.1016/j.jbusres.2018.08.009 Using a case-study based approach, this research contributes to the standardisation versus adaptation debate in global marketing. It analyses the influence of the local culture dimension reflected in consumers' comments in the Facebook platform regarding a new global technological product, the Samsung Galaxy S8/S8+, launched worldwide in 2017. Consumers' comments about this new smartphone were gathered and analysed for three cultural distinct English-speaking countries: Australia, India, and South Africa. The analysis' procedure consisted of a text mining and topic modelling approach, including sentiment classification analysis, to discern and understand consumers' responses to global brand communications. The findings indicate that cultural aspects still play a key role in consumers' reactions to the product in each country, justifying the continued need for marketing strategies that conflate pursuing economies of scale with accounting for the cultural sensitivities of demand at country level. Evidence of consumers attitudes' and behaviours' homogenisation across countries is still limited. preprint authorsversion published

In this study, we focus on understanding and mining user’s preferences and intentions via user-based aggregation in the context of a social network. Understanding preference and intention in microblog texts is more difficult and challenging than understanding such characteristics in the context of standard text. The main reason is that search history and click history are difficult to obtain due to data privacy in social networks. Meanwhile, the text is sparse, and the number of background topics in social networks is enormous. To overcome the above challenges, we explore an indirect method of user’s preference and intention understanding by leveraging a user-based aggregation topic model (UATM). Our UATM aims to mine the distributions of user’s preferences and intentions by utilizing user’s preference and intention distributions and followees’ preference and intention distributions. Furthermore, to alleviate the sparsity problem, we discriminatively model common words and topic words and incorporate a user factor into our model. We combine the recurrent neural network (RNN) and inverse document frequency (IDF) as the weight prior to learn word relationships. Moreover, to further weaken the sparsity of context, we leverage word pairs to model topics for all documents. We also propose a collapsed Gibbs sampling algorithm to infer preference and intention in our UATM. To verify the effectiveness of the proposed method, we collect a Sina Weibo dataset consisting of microblog users and their pushed content to conduct various experiments. Both qualitative and quantitative evaluations demonstrate that our proposed UATM model outperforms several state-of-the-art methods.