Alan S. Abrahams

Vehicle defect discovery from social media

A pressing need of vehicle quality management professionals is decision support for the vehicle defect discovery and classification process. In this paper, we employ text mining on a popular social medium used by vehicle enthusiasts: online discussion forums. We find that sentiment analysis, a conventional technique for consumer complaint detection, is insufficient for finding, categorizing, and prioritizing vehicle defects discussed in online forums, and we describe and evaluate a new process and decision support system for automotive defect identification and prioritization. Our findings provide managerial insights into how social media analytics can improve automotive quality management.

ExpertRank: A topic-aware expert finding algorithm for online knowledge communities

With increasing knowledge demands and limited availability of expertise and resources within organizations, professionals often rely on external sources when seeking knowledge. Online knowledge communities are Internet based virtual communities that specialize in knowledge seeking and sharing. They provide a virtual media environment where individuals with common interests seek and share knowledge across time and space. A large online community may have millions of participants who have accrued a large knowledge repository with millions of text documents. However, due to the low information quality of user-generated content, it is very challenging to develop an effective knowledge management system for facilitating knowledge seeking and sharing in online communities. Knowledge management literature suggests that effective knowledge management should make accessible not only written knowledge but also experts who are a source of information and can perform a given organizational or social function. Existing expert finding systems evaluate ones expertise based on either the contents of authored documents or ones social status within his or her knowledge community. However, very few studies consider both indicators collectively. In addition, very few studies focus on virtual communities where information quality is often poorer than that in organizational knowledge repositories. In this study we propose a novel expert finding algorithm, ExpertRank, that evaluates expertise based on both document-based relevance and ones authority in his or her knowledge community. We modify the PageRank algorithm to evaluate ones authority so that it reduces the effect of certain biasing communication behavior in online communities. We explore three different expert ranking strategies that combine document-based relevance and authority: linear combination, cascade ranking, and multiplication scaling. We evaluate ExpertRank using a popular online knowledge community. Experiments show that the proposed algorithm achieves the best performance when both document-based relevance and authority are considered.

What's buzzing in the blizzard of buzz? Automotive component isolation in social media postings

In the blizzard of social media postings, isolating what is important to a corporation is a huge challenge. In the consumer-related manufacturing industry, for instance, manufacturers and distributors are faced with an unrelenting, accumulating snow of millions of discussion forum postings. In this paper, we describe and evaluate text mining tools for categorizing this user-generated content and distilling valuable intelligence frozen in the mound of postings. Using the automotive industry as an example, we implement and tune the parameters of a text-mining model for component diagnostics from social media. Our model can automatically and accurately isolate the vehicle component that is the subject of a user discussion. The procedure described also rapidly identifies the most distinctive terms for each component category, which provides further marketing and competitive intelligence to manufacturers, distributors, service centers, and suppliers.

Automated defect discovery for dishwasher appliances from online consumer reviews

Online dishwasher reviews contain many postings relating to dishwasher defects.We assess the effectiveness of sentiment analysis for dishwasher defect discovery.We propose new smoke term dictionaries for enhancing dishwasher defect discovery.Smoke terms deliver comparable performance to the best sentiment-based technique.Dishwasher smoke terms are distinct from sentiment terms, and mainly non-emotive. Product defects can have a devastating impact on a firms sales and reputation, especially in the era of social media. The early detection of defects could not only protect consumers from financial losses, but could also mitigate financial damage to the manufacturer. Previous work in automated defect discovery has had success in the automotive, consumer electronics, and toy industries, but so far there has been no application to home appliances. In this study, we extend the text analytic framework conceived in earlier work to the discovery of underperformance in large home appliances, specifically dishwashers. We find that generic cross-domain sentiment techniques can be strongly complemented by domain-specific smoke and sparkle term lists that are highly correlated with potential defects. These findings can be highly beneficial to improving dishwasher appliance quality management methods.

Toy safety surveillance from online reviews

Toy-related injuries account for a significant number of childhood injuries and the prevention of these injuries remains a goal for regulatory agencies and manufacturers. Text-mining is an increasingly prevalent method for uncovering the significance of words using big data. This research sets out to determine the effectiveness of text-mining in uncovering potentially dangerous childrens toys. We develop a danger word list, also known as a smoke word list, from injury and recall text narratives. We then use the smoke word lists to score over one million Amazon reviews, with the top scores denoting potential safety concerns. We compare the smoke word list to conventional sentiment analysis techniques, in terms of both word overlap and effectiveness. We find that smoke word lists are highly distinct from conventional sentiment dictionaries and provide a statistically significant method for identifying safety concerns in childrens toy reviews. Our findings indicate that text-mining is, in fact, an effective method for the surveillance of safety concerns in childrens toys and could be a gateway to effective prevention of toy-product-related injuries.

Ensemble methods for advanced skier days prediction

The tourism industry has long utilized statistical and time series analysis, as well as machine learning techniques to forecast leisure activity demand. However, there has been limited research and application of ensemble methods with respect to leisure demand prediction. The research presented in this paper appears to be the first to compare the predictive power of ensemble models developed from multiple linear regression (MLR), classification and regression trees (CART) and artificial neural networks (ANN), utilizing local, regional, and national data to model skier days. This research also concentrates on skier days prediction at a micro as opposed to a macro level where most of the tourism applications of machine learning techniques have occurred. While the ANN model accuracy improvements over the MLR and CART models were expected, the significant accuracy improvements attained by the ensemble models are notable. This research extends and generalizes previous ensemble methods research by developing new models for skier days prediction using data from a ski resort in the state of Utah, United States.

Ensemble learning methods for pay-per-click campaign management

We propose using ensemble techniques to improve pay-per-click campaign management.Actual data containing pay-per-click campaign results is used for our analysis.Four base classifiers are analyzed as inputs to four ensemble modeling techniques.A MetaCost ensemble predicted the highest campaign portfolio profit. Sponsored search advertising has become a successful channel for advertisers as well as a profitable business model for the leading commercial search engines. There is an extensive sponsored search research stream regarding the classification and prediction of performance metrics such as clickthrough rate, impression rate, average results page position and conversion rate. However, there is limited research on the application of advanced data mining techniques, such as ensemble learning, to pay per click campaign classification. This research presents an in-depth analysis of sponsored search advertising campaigns by comparing the classification results from four base classification models (Naive Bayes, logistic regression, decision trees, and Support Vector Machines) with four popular ensemble learning techniques (Voting, Boot Strap Aggregation, Stacked Generalization, and MetaCost). The goal of our research is to determine whether ensemble learning techniques can predict profitable pay-per-click campaigns and hence increase the profitability of the overall portfolio of campaigns when compared to standard classifiers. We found that the ensemble learning methods were superior classifiers based on a profit per campaign evaluation criterion. This paper extends the research on applied ensemble methods with respect to sponsored search advertising.

Audience targeting by B-to-B advertisement classification: A neural network approach

As marketing communications proliferate, the ability to target the right audience for a message is of ever-increasing importance. Audience targeting practices for mass media, both in research and in industry, have tended to emphasize demographics, behavior, and other characteristics of customer groups as the bases for matching communications to audiences. These approaches overlook the opportunity to leverage the nature of advertising content, by automatically matching advertisement content to appropriate media channels and target audience. We model the semantic and sentiment content of advertisements with 103 variables. Based on these variables, a neural network classifier is used to assign advertisements to groups that represent different media channels. In its ability to classify unseen advertisements, the model outperforms the classification result generated by a random model, by 100-300%. This method also enables us to identify and describe divergent advertisement characteristics, by industry.

An Analytical Framework for Understanding Knowledge-Sharing Processes in Online Q&A Communities

Online communities have become popular knowledge sources for both individuals and organizations. Computer-mediated communication research shows that communication patterns play an important role in the collaborative efforts of online knowledge-sharing activities. Existing research is mainly focused on either user egocentric positions in communication networks or communication patterns at the community level. Very few studies examine thread-level communication and process patterns and their impacts on the effectiveness of knowledge sharing. In this study, we fill this research gap by proposing an innovative analytical framework for understanding thread-level knowledge sharing in online Q&A communities based on dialogue act theory, network analysis, and process mining. More specifically, we assign a dialogue act tag for each post in a discussion thread to capture its conversation purpose and then apply graph and process mining algorithms to examine knowledge-sharing processes. Our results, which are based on a real support forum dataset, show that the proposed analytical framework is effective in identifying important communication, conversation, and process patterns that lead to helpful knowledge sharing in online Q&A communities.

Inducing a marketing strategy for a new pet insurance company using decision trees

In this paper, we demonstrate the use of decision tree induction for the creation of a marketing strategy for a new pet insurance company, PetPlan USA. We employ both a traditional C4.5 decision tree approach, and a novel locally profit-optimal decision algorithm, called SBP, to discover the characteristics of profitable demographics for PetPlan to market to. We use publicly available data, including US census data, and veterinary clinic location data as our data sources. We evaluate our results, and give actionable recommendations for the managers of PetPlan USA. Our results indicate that entropy-based decision tree induction approaches, which focus on node purity (predominance of one category over another at each node in the tree), can produce lower profits compared to SBP, which is a novel profit-based decision tree approach.