Vocabulary Project

Lib 202-11

Kim Woolley and Kristi Mansolf

User Model

User: San JoseStateUniversity, School of Library Information Science (SLIS), students who have taken Library 202, are familiar with precoordinate indexing systems, and will be using the file for the rest of their program.

Objective: To find articles about information retrieval systems that will help the user with class projects for SLIS courses by using a controlled vocabulary that further describes the information in the article, thereby assisting them in achieving better recall and precision during information retrieval. Students will be researching aspects of retrieval systems.

Queries: To better define the User, subject areas located in the abstracts of the In Magic Database System were considered. The kinds of queries that will be conceptually formulated by the user are presumed to be those pertaining to the following subject areas:

  • Models of Information Retrieval Systems
  • User Needs
  • User Satisfaction
  • Cognitive Processes during Information Retrieval

The queries also include key points that reflect the authors’ preliminary Investigation of the In Magic Database System that was conducted in October, 2004. For this project, three sample queries were formulated to gain familiarity with the In Magic Database System, and were measured for recall and precision for the Title, Abstract and ERIC fields. The same queries will be used to measure recall and precision for the precoordinate fields created. ERIC descriptors will be used to create the strings for the precoordinate fields.

Rules:

  1. The general topic of information retrieval systems will be assumed the focus for all articles;
  2. Directly searchable items will be used to determine what aspect(s) of information retrieval systems the article looks at;
  3. General aspects of information retrieval systems may be used as both directly searchable items and modifiers;
  4. Subtopics will determine within what context the aspect is discussed;
  5. Geographic references will be reserved for the final position;
  6. Classifications of users will be reserved for the final position
  7. Rule of specific entry: Indexing will be at level of specificity of the document not broader or narrower;
  8. Broader – narrower relationships will not be used; and

9.All keywords read as: aspect of system in the context of the subtopic.

“:= in the context of”

Methodologies: The precoordinate indexing field was tested using the following three subjects: 1) User Satisfaction with Information Systems; 2) User Satisfaction with Online Catalog Systems; and 3) Cognitive Processes during Information Seeking. Each subject was considered individually, with strings entered that could be relevant to the query for each article. Determination of relevance of the query to the articles was then made, and recall and precision were calculated for each query using the “Preco” field. In addition, searches on the same subjects were conducted using the title, abstract, and ERIC fields in order to create a frame of reference.

Results:

User Satisfaction with Online Catalog Systems

Recall

Precision

For this query recall is 57 percent, with 14 articles retrieved that initially match the problem definition, out of the database of 18 articles, with 8 articles actually matching the problem definition. Precision is also 57 percent, again with 8 articles actually matching the problem definition, out of 14 relevant articles retrieved by the system.

Cognitive Processes during Information Seeking

Recall

Precision

For the query “Cognitive Processes during Information Seeking”, recall is 100 percent, with 10 articles retrieved that initially match the problem definition, out of the database of 18 articles, with 10 articles actually matching the problem definition. Precision is also 100 percent, again with the search locating 10 articles that actually match the problem definition out of 10 articlesretrieved by the system.

User Satisfaction with Information Systems

Recall

Precision

The final subject search revealed 66 percent recall and 36 percent precision from the preco field. It was difficult to find strings that exactly expressed user satisfaction which is reflected in the low precision rate. It was noticed that because multiple strings had to be used in order to cover the subject area, there was a larger query result than in the other types of searches. The query resulted in 11 articles being retrieved and only 4 relevant of 6 available in the database were retrieved.

Conclusions:

In reviewing the precoordinate index strings after completion, an observation was made that few of the strings repeated. This was not done deliberately and may be an indicator of inexperienced creators. This may also be a sign that more specific rules may have been needed. It was difficult to exactly define the articles in strings when confined to the ERIC descriptors. Often, the term options did not provide the exact term needed to truly capture the articles contents. For example, by analyzing the strings that were not relevant to the second search, “User Satisfaction with Online Catalog Systems”, the strings could have been related to the query, but upon closer examination, were not. By attempting to use all of the ERIC descriptors to design the Preco field, some of the ERIC descriptors, although relevant to the article, did not fit the query.

The strings of the Preco field enabled the User to focus the search more, reducing the gap between recall and precision. When using the In Magic Database System to design the Preco field, the strings created did better reflect the content of the articles than the ERIC descriptors alone. This experiment did reveal the benefits of using precoordinate strings. With experience this system could definitely benefit SJSU students search for information. The strings provided a basis for defining the subject searches. In addition, the connection of relevant terms provided a context giving users more accurate expectations of the articles’ contents. Controlled vocabularies and precoordinate search fields are obviously an area that require a great deal of experience and expertise in order to master. However, our experiment gave us a greater understanding of how these systems work and a greater respect for their creators.

Libr 202 Strings for Vocab Project

Article 01 – Title: “Models of User Satisfaction: Understanding False Positives”

Models: User Satisfaction

User Satisfaction: online searching

User Satisfaction: Research Methodology

Article 02 – Title: “The Design of Browsing and Berrypicking Techniques for the Online Search Interface”

Models: Online Searching

Models: Search Behavior

User Needs: System Development

User Needs: User Cordial Interface

Article 06 – Title: “Expertise, Task Complexity, and Artifician Intelligence: A Conceptual Framework”

Artificial Intelligence: Cognitive Models

Artificial Intelligence: Computer System Design

Artificial Intelligence: Conceptual Models

Artificial Intelligence: Difficulty Level

Artificial Intelligence: Systems Analysis

Cognitive Models: Artificial Intelligence

Expertise: Information Systems

Information Systems: Systems Analysis—Difficulty Level

Information Systems: Systems Analysis—Expertise

Users [Information]: Conceptual Models

Users [Information]: Cognitive Models

Article 08 – Title: “Retrieval by Reformulation in Two Library Catalogs: Toward a Cognitive Model of Searching Behavior”

Cognitive Processes: Card Catalog

Cognitive Processes: Online Catalogs

Cognitive Processes: Search Strategies

Higher Education: Academic Libraries

Online Catalogs: Search Behavior

Online Catalogs: Search Strategies

Article 09 – Title: “A Cognitive Process Model of Document Indexing”

Conceptual Approach: Scanning

Cognitive Models: Cognitive Processes

Cognitive Processes: Abstracting

Cognitive Processes: Indexing

Cognitive Processes: Classification

Model: Abstracting

Model: Indexing

Model: Classification

Reading Comprehension: Long Term Memory

Reading Comprehension: Scanning

Reading Comprehension: Short Term Memory

Abstract 14 – Title: “Windows into the Search Process: an Inquiry into Dimensions of Online Information Retrieval”

Cognitive Models: Search Strategies

Contextual Analysis: Information Retrieval

Contextual Analysis: Information Technology

Contextual Analysis: Online Systems

Futures [of Society]: Information Retrieval

Futures [of Society]: Search Strategies

Information Retrieval: Search Strategies

Models: Cognitive Models

Search Strategies: Cognitive Models

Users: Research Needs

Abstract 20 – Title: “Organizational Factors in Human Memory:

Implications for Library Organization and Access Systems”

Cognitive Processes: Classification

Cognitive Processes: Organization

Cognitive Processes: Psychological Studies

Information Systems: Classification

Information Systems: Memory

Information Systems: Organization

Memory: Psychological Studies

Abstract 23 – Title: “Information Processing Models of Cognition”

Cognitive Development: Concept Formation

Cognitive Processes: Computers—Simulation

Cognitive Processes: Cognitive Development

Cognitive Processes: Concept Formation

Cognitive Processes: Information Processing

Cognitive Processes: Memory

Cognitive Processes: Pattern Recognition

Cognitive Processes: Problem Solving

Computers: Simulation

Memory: Pattern Recognition

Memory: Problem Solving

Models: Cognitive Processes

Abstract 25 – Title: “Term Relevance Feedback and Mediated Database

Searching: Implications for Information Retrieval Practice and Systems

Design”

Abstract 28 – Title: “Beyond Topical Relevance: Document Selection

Behavior of Real Users of IR Systems”

Artificial Intelligence: Cognitive Models

Artificial Intelligence: Computer System Design

Artificial Intelligence: Conceptual Models

Artificial Intelligence: Document Handling

Cognitive Model: Computer System Design

Cognitive Model: Online Searching

Decision Making: Relevance

Document Handling: Cognitive Model

Document Handling: Decision Making

Document Handling: Evaluation Criteria

Document Handling: Value Judgment

Information Retrieval: Document Handling

Models: Decision Making

1