EXPRESS: Attribute Embedding: Learning Hierarchical Representations of Product Attributes from Consumer Reviews

Sales, product design, and engineering teams benefit immensely from better understanding customer perspectives. How do customers combine a product’s technical specifications (i.e., engineered attributes) to form abstract product benefits (i.e., meta-attributes)? To address this question, the authors use machine learning and natural language processing to develop a methodological framework that extracts a hierarchy of product attributes based on contextual information of how attributes are expressed in consumer reviews. The attribute hierarchy reveals linkages between engineered attributes and meta-attributes within a product category, enabling flexible sentiment analysis that can identify how meta-attributes are received by consumers, and which engineered attributes are main drivers. The framework can guide managers to monitor only portions of review content that are relevant to specific attributes. Moreover, managers can compare products within and between brands, where different names and attribute combinations are often associated with similar benefits. The authors apply the framework to the tablet computer category to generate dashboards and perceptual maps, and provide validations of the attribute hierarchy using both primary and secondary data. Resultant insights allow the exploration of substantive questions, such as how successive generations of iPads were improved by Apple, and why HP and Toshiba discontinued their tablet product lines.

A Grey Multi-Level Evaluation of Industrial Park Ecology Based on a Coefficient of Variation-Attribute Hierarchy Model

China entered a new era, and the construction of an ecological civilization and green development has been raised to a new strategic height. As the lifeblood of the national economy, industrial parks significantly contribute to economic growth. However, they also generate significant pollution, damaging the ecological environment. It is urgent to ecologically transform traditional industrial parks. This requires identifying methods to correctly and objectively evaluate the ecological level of industrial parks, and provide ecological construction proposals for the government and industrial parks. In this study, the comprehensive evaluation weight was determined by introducing a variation coefficient and an Attribute Hierarchy Model (AHM). The ecological level of four representative eco-industrial parks was then quantitatively evaluated using a grey multi-level evaluation method. The ecological construction level of the four industrial parks was as follows. The Tianjin Economic-Technological Development Area (TEDA) was rated at a “very good” level; and the Suzhou industrial park, Dalian economic and technological development zone, and Fushun mining group were rated at a “good” level. Six dimensions were studied. Of these, policy management had the highest weight, and the total weight of policy management and economic development approached 50%. The result shows that industrial parks can attract innovative enterprises and talents through the policy guidance of local government to improve the level of green innovation technology and cleaner production technology. Then, the ecological level of the industrial parks will be improved. This study enriched the theory and practice of ecological evaluation of industrial parks and provided a reference for the ecological construction of traditional industrial parks.

Devising an Algorithm for Election Prediction Using Survey of Voters Opinions

The purpose of this research study is to analyze how we use voter polls to predict elections and to design an algorithm to predict elections. We propose a method of prediction based on learning algorithm to determine the political profile of a voter group by obtaining a linear hierarchy on the attributes that weights the number of instances that are more relevant. Our process starts with opinion survey collected directly from the target group of voters. Having a linear attribute hierarchy that expresses the political preferences of voters allows the application of a holistic approach to distribute the potential number of votes among the parties involved. We applied our electoral outlook model in the Kosovo election case study in from February 2021. The devised algorithmic model may also be applied to other situations. Data analysis not only provides new analysis opportunities, but also faces many challenges. In our case, we listed the limitations of the research. The research attempts to promote the implementation of the algorithm by extending the processing of the information generated by the learning algorithm to improve the prediction of elections and winning parties. Discussed of all data analysis challenges, and present, discuss, and argue insights.

Multi-Criteria Evaluation of Best Available Treatment Technology for Waste Lead-Acid Battery: The Case of China

Improper waste lead-acid battery (LAB) disposal not only damages the environment, but also leads to potential safety hazards. Given that waste best available treatment technology (BATT) plays a major role in environmental protection, pertinent research has largely focused on evaluating typical recycling technologies and recommending the BATT for waste LABs. First the evaluation indicators were selected based on the analysis of main factors affecting the pollution control of waste LAB treatment. The relative weights of each indicator were determined via the Delphi-attribute hierarchy model (AHM) in the second step. To determine the BATT, the attributive mathematics theory was adopted to calculate the attribute measure of single and multiple indices. Then, five recycling technologies commonly used in the secondary lead industry were estimated using the proposed evaluation system, and the feasibility of the recommended BATT was preliminarily verified. The results indicated that mixed smelting technology (MST), pre-desulfurization and multi-chamber smelting technology (PD-MCST), and direct smelting technology (DST) were found to perform well and were therefore deemed optimal for waste LAB disposal at this stage. The validation study showed that the DST can meet the requirements of pollution control, which is consistent with the evaluation results.

Complete Q-Matrices in General Attribute Structure Models

In cognitive diagnostic assessment a property of the Q-matrix, usually referred to as completeness, warrants that the cognitive attributes underlying the observed behavior can be assessed uniquely. Characterizations of completeness were first derived under the assumption of independent attributes, and are currently under investigation for interdependent attributes. The dominant approach considers so-called attribute hierarchies, which are conceptualized through a partial order on the set of attributes. The present paper corrects and extends previously published results on this issue obtained for conjunctive attribute hierarchy models. Drawing upon results from knowledge structure theory it provides novel sufficient and necessary conditions for completeness of the $Q$-matrix, not only for conjunctive models on attribute hierarchies, but also on more general attribute structures.