Hsiao Tieh Pu

A comparative analysis of web image and textual queries

Purpose – This study aims to examine the differences between web image and textual queries.Design/methodology/approach – A large number of web queries from image and textual search engines were analysed and compared based on their factual characteristics, query types, and search interests.Findings – Useful results include the findings that web users tend to input short queries when searching for visual or textual information; that image requests have more zero hits and higher specificity, and contain more refined queries; that web image requests are more focused than textual requests on some popular search interests, and that the variety of textual queries is greater than that of image requests.Originality/value – This study provides results that may enhance ones understanding of web‐searching behaviour and the inherent implications for the improvement of current web image retrieval systems.

An analysis of Web image queries for search

This study examines the differences between Web image and textual queries, and attempts to develop an analytic model to investigate their implications for Web image retrieval systems. A large number of Web queries from image and textual search engines were analyzed and compared based on their factual characteristics, query types, and search interests. A feasible analytic model employing the concepts of uniqueness and refinement was adapted to categorize query types and analyze the characteristics of failed queries. Useful results include the findings that image requests may have higher specificity and contain more refined queries (especially among failed queries), and that the queries were refined more by interpretive attributes than by reactive and perceptual attributes. Current text retrieval technology is not capable of fulfilling such complex image requests. It is suggested that there is a need to increase the number of appropriate annotations for Web images and to utilize more advanced retrieval techniques for more effective Web image searching. Few previous large-scale studies have investigated visual information retrieval using image search engines. Thus, this study provides results that might enhance our understanding of Web image searching behavior and suggests implications for the improvement of current Web image search engines.

Enriching user‐oriented class associations for library classification schemes

This paper explores the possibility of adding user‐oriented class associations to hierarchical library classification schemes. Some highly associated classes not grouped in the same subject hierarchies, yet relevant to users’ knowledge, are automatically obtained by analyzing a two‐year log of book circulation records from a university library in Taiwan. The library uses the Chinese Decimal Classification scheme, which has similar structure and notation to the Dewey Decimal Classification. Methods, from both collaborative filtering and information retrieval research, were employed and their performance compared based on similarity estimation of classes. The results show that classification schemes can, therefore, be made more adaptable to changes of users and the uses of different library collections by analyzing the circulation patterns of similar users. Limitations of the methods and implications for applications are also discussed.

Auto-construction of a live thesaurus from search term logs for interactive Web search (poster session)

The purpose of this paper is to present an on-going research that is intended to construct a live thesaurus directly from search term logs of real-world search engines. Such a thesaurus designed can contain representative search terms, their frequency in use, the corresponding subject categories, the associated and relevant terms, and the hot visiting Web sites/pages the search terms may reach.

User Navigation for Online Video Leisure: An Exploratory Study

This paper describes the navigation tendencies of the users of online video platforms for leisure purposes. Ten participants with web video leisure experience were recruited for a month-long study. The study employed multiple methods including questionnaires, screen logging, diaries and interviews. The analysis focused on navigation time, navigation moves, and system functions used. The preliminary results show that the average navigation time for each session was 28.43 minutes, and nearly 80% of the time was spent in watching videos vs. 20% in interacting with the system; each session included 6.8 clicks to switch windows, 5.22 clicks on related videos, and 0.68 clicks on related textual information; in each session, the number of basic system functions used averaged 18.74 compared with an average of fewer than 1 use of an advanced function. Overall, the participants concentrated mostly on the videos, though some effort was spent on other online leisure activities in parallel, and the participants showed great interest in exploring unexpected videos.

User navigation behavior of a selective dissemination of web information service

Nowadays it becomes a challenge for a user to be able to both effectively filter information and proactively grasp new information on the web. To this end, Selective Dissemination of Web Information (SDWI) service in this study is conceptualized as a new way of information delivery. Using RSS service as a test platform, the study attempts to explore how users navigate the large amount of information from RSS. The study used screen logging and interview methods. The preliminary findings show that the participants demonstrate a quick screening reading style, since near 70% time and 90% moving were related to filtering and browsing contents in RSS readers, while only 30% time and 10% moving were concerned with reading source pages. The study further categorizes five different navigation patterns according to the depth of reading and navigation. The paper concludes that SDWI would receive more attention in information related research and applications.

Annotating text segments using a web-based categorization approach

Conventional automatic text annotation tools mostly extract named entities from texts and annotate them with information about persons, locations, and dates, etc. Such kind of entity type information, however, is insufficient for machines to understand the context or facts contained in the texts. This paper presents a general text categorization approach to categorize text segments into broader subject categories, such as categorizing a text string into a category of paper title in Mathematics or a category of conference name in Computer Science. Experimental results confirm its wide applicability to various digital library applications.

A query analytic model for image retrieval

Searching digital images on a networked environment is rapidly growing. Despite recent advances in image retrieval technologies, high-precision and robust solutions remain hampered by limits to knowledge about user issues associated with image retrieval. This paper examines a large number of queries from a Web image search engine, and attempts to develop an analytic model to investigate their implications for image retrieval technologies. The model employs the concepts of uniqueness and refinement to categorize successful and failed queries. The results show that image requests have a higher specificity and may often contain queries refined by interpretive, reactive, and perceptual attributes. Based on the proposed model, the study further investigates feasible technical solutions integrating both content-based and concept-based technologies to deal with real image query types. The initial study has provided useful results that enhance the understanding of digital image searching and suggests implications for the improvement of image retrieval systems.