CS/INFO 430 Information Retrieval Lecture 22 Usability 1 1 Course Administration 2 Browsing: The Human in the Loop Return objects Return hits Browse repository Search index 3 Web Search: Browsing Users give queries of 2 to 4 words Most users click only on the first few results; few go beyond the fold on the first page 80% of users, use search engine to find sites search to find site browse to find information Amil Singhal, Google, 2004 4 Browsing in Information Space Starting point x x x x x x x x x x x x x x Effectiveness depends on (a) Starting point (b) Effective feedback (c) Convenience 5 Convenience of Browsing Rapid access to materials Physical objects • Library or large private collection • Similar items stored close together (classification) Online information • Rapid delivery to desktop good system performance no administrative delays (authentication) Human skills and knowledge augment and extend the automatic methods of searching 6 Convenience of Browsing If the documents are accessible online, user can browse content. • This can compensate for weaknesses in the underlying search system, e.g., the difficulty of indexing Web documents Otherwise, the user can browse substitutes, e.g., catalog records, subject hierarchies, etc. • This puts heavy demands on the precision/recall of the underlying search system 7 Browse: Catalog Record QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 8 Browsing the Content of Indexes Show the users the terms that occur in indexes, such as subject headings. Example: Library of Congress:American Memory http://memory.loc.gov/ 9 Hierarchical browsing Level 0 Level 1 Level 2 10 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Alphabetic list http://nsdl.org/ 11 Broad categories QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 12 QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 13 Subject headings grouped QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 14 Subject headings used in index Browsing by Filtering and Sorting Filters allow users to reject categories of information. Sorting by various criteria allows users to organize information for rapid scanning Example: Research Libraries Group Cultural Materials http://cmi.rlg.org/ 15 Browse everything QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 16 Filter "New York" QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 17 Sort "date" QuickTime™ and a and a TIFFQuickTime™ (LZW) decompressor (LZW) decompressor areTIFF needed to see this picture. are needed to see this picture. 18 Designing the Search Page Making Decisions • Overall organization: – Spacious or cramped – Division of functionality to different pages – Positioning components in the interface – Emphasizing parts of the interface • Query insertion: insert text string or fill in text boxes • Interactivity of search results • Performance requirements 19 Google 20 Spacious organization AltaVista Division of functionality to different pages 21 ACM Digital Library Emphasized components 22 ACM Digital Library advance Different searchquery insertion ways 23 Yahoo! 24 cramped organization The Old Yahoo! Interface 25 The Yahoo! Interface The Yahoo interface is cluttered and unattractive, yet Yahoo is one of the most successful of all web sites. Why is this interface successful? • Very many branches from a single web page saves the need for hierarchy of menus. • Simple html markup ensures that the page renders quickly and accurately on all browsers. • Slow changes over the years means that users are familiar with it. http://www.yahoo.com/ 26 27 Return Hits: Snippets • A snippet is a short record that a search system returns to describe and link to a hit. • Example: Web search “Nielsen evaluation heuristics” Heuristic Evaluation ... Jacob Nielsen's Online Writings on Heuristic Evaluation. How to conduct a heuristic evaluation; A list of ten recommended heuristics for usable interface design ... www.useit.com/papers/heuristic/ - 5k - Cached - Similar pages 28 Usability of Search: Snippets Choices in designing snippets: • Dynamic (generated from query + document) or precomputed (from document only) • Content only or with related information (e.g., subject hierarchies) • Highlighting of search terms • Length of snippet v. number on page User must understand why the hit was returned 29 Dynamic Return Hits 30 Dynamic snippets Precomputed Return HitsPrecomputed snippets 31 Pre-computed Snippets In general dynamic snippets are superior because they fit the user's expectations, but they can fail badly. Example: Web search "brown topeka kansas" QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Legal Information Institute Brown v. Board of Education, 347 U.S. 483 (1954) (USSC+) 1. Syllabus , 2. Full Decision , 3. Syllabus & Opinions Only... www2.law.cornell.edu/cgi-bin/foliocgi.exe/... 32 Dynamic Snippets QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. Legal Information Institute www2.law.cornell.edu/.../doc/%7B@1%7D/ hit_headings/words=4/hits_only - 2k - Oct 27, 2003 - Cached -Similar pages DOC BodyPage ... Case Information. Brown v. Board of Education of Topeka . No. 1. ... APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF KANSAS [*]. Syllabus. ... www2.law.cornell.edu/.../doc/%7Bt26262%7D/ pageitems=%7Bbody%7D/hit_headings/words=4 - 13k - Cached Similar pages 33 Pre-computed Snippets QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 34 Dynamic Snippets with Pre-computed Summary QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 35 Dynamic Snippets with Pre-computed Summary Pre-computer summary, with space for dynamic snippet QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 36 Dynamic Snippets with Pre-computed Summary Complete record with dynamic snippet QuickTime™ and a TIFF (LZW) decompressor are needed to see this picture. 37