基于CP-nets 的偏好感知交互式遗传算法及其个性化搜索

针对用户显式评价导致用户疲劳,进而限制交互式遗传算法搜索性能的问题,研究基于用户交互行为和条件偏好网络(CP-nets)的隐式评价模式的交互式遗传算法,并将其应用于图书商品个性化搜索。首先,给出用户交互行为的数学描述,建立基于用户少量交互行为的条件偏好网络模型以拟合用户偏好;然后,利用CP-nets模型估计用户对进化个体的评价值,实施进化操作以帮助用户尽快找到满意解。在个性化搜索中的应用验证了所提出算法的有效性。     

足球视频搜索引擎中的用户偏好挖掘

目的：互联网信息量的急速增长使得人们需要花费大量时间从搜索引擎召回的结果中浏览自身感兴趣的内容,结合用户的搜索日志信息和社交平台信息,提出一种分层的实时偏好挖掘模型,为用户提供个性化搜索服务。方法：在系统分析偏好挖掘的国内外研究现状的基础上,针对足球视频,提出一种分层权重无向图（Hierarchical Weighted Undirected Graph,HWUG）用户偏好模型,充分考虑了用户偏好之间的关联信息,通过获取用户足球领域的显式和隐式反馈信息,提取反馈信息中的偏好标签和偏好动作,并引入时间衰减因子,实现了用户足球偏好的实时计算。结果：算法已经应用在在搜球网（www.findball.net）的个性化检索结果排序和视频推荐上,并已经取得了很好的效果。结论：实验结果表明,结合特定领域的知识,基于分层无向权重图模型的偏好挖掘算法能更准确和实时反映用户的足球偏好。     

Intelligent front-ends and product models

Research efforts in two separate areas, intelligent user interfaces and product data models have in the past been carried out almost totally independently from each other. It would, however, seem that product models could be integrated into Intelligent Front-Ends and that results from Intelligent Front-End research could be used for building the user interfaces to database information stored according to product model schemas.The paper starts with an introduction to data base theory and then moves on to present general features of product models. The presentation is on a general level but the examples are picked from the building domain. The international effort to define a standardized basic product model framework, the PDES/STEP project, and the data modeling methods used in it are also described. Work with prototype interfaces to product data models for the building domain is described as an example of the role Intelligent Front-Ends could play in product model based applications. The paper ends with a short discussion of the role product models could have in Intelligent Front-Ends.     

Large scale product search with spatial quantization and deep ranking

Multimedia Tools and Applications - Product image search aims to retrieve similar product images based on a query image. While deep learning based features work well in retrieving images of the...     

Interactive ads recommendation with contextual search on product topic space

The rapid popularization of various online media services have attracted large amounts of consumers and shown us a large potential market of video advertising. In this paper, we propose interactive service recommendation based on ad concept hierarchy and contextual search. Instead of traditional ODP (Open Directory Project) based approach, we built a ad domain based concept hierarchy to make the most of the product details over the e-commerce sites. Firstly, we capture the summarization images related to the advertising product in the video content and search visually similar product images from the built product image database. Then, we aggregate the visual tags and textual tags with K-line clustering. Finally, we map them to the product concept space and make keywords suggestion, and users can interactively select keyframes or keywords to personalize their intentions by textual re-search. Experiments and comparison show that the system can accurately provide effective advertising suggestions.     

Exploratory product search using top-k join queries

Given a relation that contains main products and a set of relations corresponding to accessory products that can be combined with a main product, the Exploratory Top-k Join query retrieves the k best combinations of main and accessory products based on user preferences. As a result, the user is presented with a set of k combinations of distinct main products, where a main product is combined with accessory products only if the combination has a better score than the single main product. We model this problem as a rank-join problem, where each combination is represented by a tuple from the main relation and a set of tuples from (some of) the accessory relations. The nature of the problem is challenging because the inclusion of accessory products is not predefined by the user, but instead all potential combinations (joins) are explored during query processing in order to identify the highest scoring combinations. Existing approaches cannot be directly applied to this problem, as they are designed for joining a predefined set of relations. In this paper, we present algorithms for processing exploratory top-k joins that adopt the pull-bound framework for rank-join processing. We introduce a novel algorithm (XRJN) which employs a more efficient bounding scheme and allows earlier termination of query processing. We also provide theoretical guarantees on the performance of this algorithm, by proving that XRJN is instance-optimal. In addition, we consider a pulling strategy that boosts the performance of query processing even further. Finally, we conduct a detailed experimental study that demonstrates the efficiency of the proposed algorithms in various setups.     

用于图像购物搜索的局部匹配算法

图像匹配是图像购物搜索的一个关键问题。针对现有的图像购物搜索网站利用底层特征无法进行局部匹配的问题,提出一种基于局部匹配的图像购物搜索方法,该方法允许用户自定义感兴趣区域,并利用SIFT（Scale Invariant Feature Transform）描述子提取该区域的视觉特征进行匹配。实验结果表明,该方法取得了较理想的预期效果,更好地满足了用户的实际购物搜索需求。     

An information-theoretic approach to consumer preference and product planning

Marketing managers should develop various marketing approaches in order to satisfy the needs of consumers. It is important for industries to succeed to analyze the data from consumers and to apply the quantitative analyses to marketing research. In this paper, a procedure for measuring the preferences weights of consumers for attribute levels of prototype products is proposed using paired comparison with scores. An information-theoretic procedure for measuring the consumer preference for each prototype product and the importance or utility of each attribute is presented, considering the preference weights for attribute levels as prior information. Furthermore, optimal product planning to maximize the total expected profit is discussed through numerical experiments.     

Effective product quantization-based indexing for nearest neighbor search

Multimedia Tools and Applications - Product quantization is a widely used lossy compression technique that can generate high quantization levels by a compact codebook set. It has been conducted in...     

Intelligent search in social communities of smartphone users

Social communities of smartphone users have recently gained significant interest due to their wide social penetration. The applications in this domain, however, currently rely on centralized or cloud-like architectures for data sharing and searching tasks, introducing both data-disclosure and performance concerns. In this paper, we present a distributed search architecture for intelligent search of objects in a mobile social community. Our framework, coined SmartOpt, is founded on an in-situ data storage model, where captured objects remain local on smartphones and searches then take place over an intelligent multi-objective lookup structure we compute dynamically. Our MO-QRT structure optimizes several conflicting objectives, using a multi-objective evolutionary algorithm that calculates a diverse set of high quality non-dominated solutions in a single run. Then a decision-making subsystem is utilized to tune the retrieval preferences of the query user. We assess our ideas both using trace-driven experiments with mobility and social patterns derived by Microsoft’s GeoLife project, DBLP and Pics ‘n’ Trails but also using our real Android SmartP2P (http://smartp2p.cs.ucy.ac.cy/) system deployed over our SmartLab (http://smartlab.cs.ucy.ac.cy/) testbed of 40+ smartphones. Our study reveals that SmartOpt yields high query recall rates of 95 %, with one order of magnitude less time and two orders of magnitude less energy than its competitors.     

Intelligent search agents using web-driven natural-language explanatory dialogs

A computational linguistics approach for Web-based cooperative dialogs focuses on the user's requests by automatically generating language. driven interactions that take into account the context, user feedback, and the initial search's results.     

Affective preference measurement of product appearance based on event-related potentials

The product appearance plays an important role in users’ purchase decision, while the emotions elicited by the product appearance are subtle, low intensity and not easy to obtain. Event-related potentials (ERPs) can reflect people’s psychological activities and effectively identify the characteristics of brain’s affective process. In the present study, the affective preference measurement of humidifier appearance was conducted based on the theory of affective design and event-related potentials. Firstly, the collected humidifier pictures were screened to be the experimental stimuli through multidimensional scaling analysis and cluster analysis. During the affective preference measurement experiment, the participants were asked to judge the affective preference level (liked vs. neutral vs. disliked) of humidifier pictures with the mouse button clicking (left, middle, right), and the electroencephalograph signals were recorded at the same time. ERPs results showed that the frontal and central N1 and frontal, frontal–central and central late positive potential can be taken as the indexes to measure participants’ affective preference of humidifier appearance. In conclusion, the participants’ affective preference level of humidifier appearance can be measured by ERPs. The experiment results can be used to guide product design, by measuring the brain activities of different prototypes at the phase of product development we can roughly infer which one is the user preferred or not preferred.     

Crowd intelligence: Analyzing online product reviews for preference measurement

The proliferation of product review websites produces a large, publicly accessible information resource for firms that seek to understand consumers’ preferences. To facilitate product design or improvement, we propose a novel econometric preference measurement model, the modified ordered choice model (MOCM), to extract aggregate consumer preferences from online product reviews. Moreover, to categorize customer requirements on the basis of the aggregate consumer preferences estimated by the MOCM model, we extend the Kano model and propose a marginal effect-based Kano model (MEKM). We empirically evaluate the effectiveness of the proposed MOCM model and demonstrate the utility of the proposed MEKM model.     

Intelligent sensory evaluation of industrial products for exploiting consumer's preference

This paper presents an intelligent technique based method for analyzing and interpreting sensory data provided by consumers and experts in order to predict consumer's preference on industrial products. In this study, we first introduce the concept of equilibrium consumption for estimating consumer's data variation between two different samples. This method can effectively solve the multi-peaks problem and the order correlation problem existing in consumer's data. According to the degree of consistency of data variation, we define the dissimilarities or distances between different evaluation terms used by consumers. Based on these dissimilarities, we develop a procedure for interpreting consumer's terms using normalized expert's evaluation terms. This procedure can also be used for predicting consumer's evaluation and preference from expert's evaluation on a collection of industrial products. The effectiveness of this method has been validated in one collection of cosmetic protection creams.     

Evaluating product search and recommender systems for E-commerce environments

Online systems that help users select the most preferential item from a large electronic catalog are known as product search and recommender systems. Evaluation of various proposed technologies is essential for further development in this area. This paper describes the design and implementation of two user studies in which a particular product search tool, known as example critiquing, was evaluated against a chosen baseline model. The results confirm that example critiquing significantly reduces users’ task time and error rate while increasing decision accuracy. Additionally, the results of the second user study show that a particular implementation of example critiquing also made users more confident about their choices. The main contribution is that through these two user studies, an evaluation framework of three criteria was successfully identified, which can be used for evaluating general product search and recommender systems in E-commerce environments. These two experiments and the actual procedures also shed light on some of the most important issues which need to be considered for evaluating such tools, such as the preparation of materials for evaluation, user task design, the context of evaluation, the criteria, the measures and the methodology of result analyses.     

Optimal dual-channel supply chain configuration for product categories with different customer preference of online channel

This study addresses the supply chain configuration problem of manufacturer faced with multiple channel choices. We investigate the optimal combination of traditional and online sales channels for different product categories with differing customer preference of the online channel. We have considered three distinct dual-channel supply chain configurations comprising of a manufacturer, a traditional channel and an online channel viz. retailer–e-tailer; company store–e-tailer; and retailer–e-marketplace. As the manufacturer will face a quantity allocation decision between traditional and online channels, the competition is modeled using Cournot model with the manufacturer as the Stackelberg leader. Values of optimal order quantity and price for the different entities have been established. Influence of variation in customer preference of online channel on the optimal policies is also examined. We find that the manufacturer prefers retailer–e-marketplace configuration for products with high customer preference of online channel and company store–e-tailer configuration for products with low customer preference of online channel. In addition, we find that the retailer–e-tailer configuration is dominated by company store–e-tailer and retailer–e-marketplace configuration irrespective of the product’s customer preference of online channel.     

Using online search data to forecast new product sales

Search engines are rapidly emerging to be the “go-to” sites for consumers to learn more about a product, concept or a term of interest, irrespective of the initial channel in which the interest originated — text, radio, TV, multi-media channels, word of mouth, etc. In this paper we argue that data on the search terms used by consumers can provide valuable measures and indicators of consumer interest in a product, concept or a term. Such data can be particularly valuable to managers in gauging potential product interest in a new product launch context or consumption interest in the post-release context. Based on this premise, we develop a model of pre-launch search activity and link the pre-launch search behavior and product characteristics to early sales of the product, thus providing a useful forecasting tool. Applying the model in the context of motion pictures, we find that search term usage follows rather predictable patterns in the pre-launch and post-launch periods and the model provides significant power in forecasting release week sales as a function of pre-release search activity. With advertising data included in the model, we find that the pre-release search data offers additional explanatory and forecasting power, thus highlighting the ability of the search data to capture other factors, such as possibly word-of-mouth, in impacting early sales. We offer specific insights into how managers can use search volume data and the model to plan their new product release.     

基于知识库系统的中文智能搜索引擎

随着信息技术的发展和信息量的增长,传统的搜索引擎技术日益不能满足用户信息查询的需要.目前.搜索引擎技术与人工智能(AI)技术的结合已经成为网络信息搜索的关键技术与核心思想.一种基于知识库系统的智能搜索引擎技术已成为当前研究的热点.主要介绍基于知识库系统的中文智能搜索引擎及其实现技术,以及中文智能搜索引擎的主要发展方向.     

Optimized high order product quantization for approximate nearest neighbors search

Product quantization is now considered as an effective approach to solve the approximate nearest neighbor(ANN)search.A collection of derivative algorithms have been developed.However,the current techniques ignore the intrinsic high order structures of data,which usually contain helpful information for improving the computational precision.In this paper,aiming at the complex structure of high order data,we design an optimized technique,called optimized high order product quantization(O-HOPQ)for ANN search.In O-HOPQ,we incorporate the high order structures of the data into the process of designing a more effective subspace decomposition way.As a result,spatial adjacent elements in the high order data space are grouped into the same subspace.Then,O-HOPQ generates its spatial structured codebook,by optimizing the quantization distortion.Starting from the structured codebook,the global optimum quantizers can be obtained effectively and efficiently.Experimental results show that appropriate utilization of the potential information that exists in the complex structure of high order data will result in significant improvements to the performance of the product quantizers.Besides,the high order structure based approaches are effective to the scenario where the data have intrinsic complex structures.     

Tender calls search using a procurement product named entity recogniser

A product search service in an e-Procurement Marketplace can help the suppliers to find the best suitable tenders according to their products. Various possible ways to define and specify a product by different companies make it difficult to match a tender as a product request with the similar products offered by the suppliers. Semantic search engines try to overcome this problem by understanding the intent and contextual meaning of the words within a search domain. A fundamental part of such search engines can be a named entity recogniser that extracts desired searchable elements from the search context.This paper develops a recogniser that can extract “Procurement Product” mentions from tenders and other procurement documents. A self-learning approach has been adopted in order to train the model for extracting product mentions. The proposed approach uses already known product mentions in tenders as the training data to train the model and then use the trained model to recognize the product mentions from other tenders. The accuracy of the model has been tested evaluated using tenders that have been published in public procurement e-marketplaces. The results show that the proposed approach achieved high values of precision and recall in different test datasets. The recogniser can be used as the search element extractor for semantic search in procurement e-marketplaces. Therefore, the improvement of search performance by using the recogniser is also tested in finding tenders from different public procurement resources. The results show the semantic search process which uses the recogniser improves the search precision by about 25%.