Lecture Overview Web 2.0, Tagging, Multimedia, • Introduction to Web 2.0 • Overview of Tagging Systems Folksonomies, Lecture, – Overview of tagging Important, Must Attend, … – Design and attributes Martin Halvey – Case Studies • Other Web 2.0 Technologies • Conclusions Web 2.0 Definition Web 2.0 Definition “Web 2.0 is the business revolution in “An idea in people's heads rather than a the computer industry caused by the reality. It’s actually an idea that the move to the Internet as platform, and an reciprocity between the user and the attempt to understand the rules for provider is what's emphasised. In other success on that new platform” words, genuine interactivity if you like, simply because people can upload as well as download” Tim O'Reilly (2006-12-10). Web 2.0 Compact Definition: Trying Again Stephen Fry: Web 2.0 1
Web 2.0 Definition Web 2.0 “piece of jargon” “Participatory Web” “nobody really knows what it means” “if Web 2.0 for you is blogs and wikis, then that is people to people. But that Bart Decrem (2006-06-13). Introducing was what the Web was supposed to be Flock Beta 1. Flock official blog all along” -Tim Berners-Lee Web 2.0 Web 2.0 Feature/Techniques • Allow users to do more than retrieve • Rich Internet Application techniques, AJAX information • Semantically valid XHTML and HTML • "Network as platform" computing, allowing • REST and/or XML- and/or JSON-based APIs users to run software applications entirely • Cascading Style Sheets through a browser • RSS or Atom feeds • In contrast to systems which categorise users • Mash Ups into roles with varying degrees of functionality • Weblog publishing • Re-use of existing technologies, some new • Wiki or Forum • Social Networking 2
Tagging Folksonomies • A tag is a (relevant) keyword or term • A folksonomy is a user generated taxonomy used to categorize and retrieve web content associated with or assigned to a piece such as Web pages, photographs and Web of information (e.g. a picture, article, or links, using open-ended labels called tags video clip), thus describing the item and • Intended to make a body of information enabling keyword-based classification increasingly easy to search, discover, and of information. navigate over time • Usually chosen informally and • Normally used online but they can arise in a personally by item author/creator or by number of other contexts its consumer/viewers/community Folksonomies and Semantic Folksonomies Web • Well-developed folksonomy is ideally • The Semantic Web is an evolving extension of the WWW in which content is expressed accessible as a shared vocabulary that not only in a format that can be read and is both created by, and familiar to, its used by automated tools, as well as natural primary users language • Folksonomy tools are not part of the • Folksonomies can be used in conjunction with WWW protocols semantic web technologies to provide rich descriptions, but not quite yet. • Arise where special provisions are made • However metadata from folksonomies is not consistent or reliable • Particularly useful when no other text is available 3
Bridging the Semantic Gap Example Tagging Systems • The difference between low-level data • Del.ioco.us (http://del.ioco.us) representation of multimedia and the • CiteULike (http://www.citelike.org) higher level concepts users associate • Flickr (http://www.flickr.com) with the same multimedia • YouTube (http://www.youtube.com) • Providing annotations can alleviate this • Last.fm (http://www.last.fm) problem • Technorati (http://www.technorati.com) • Need to improve annotations to overcome this problem • ESP Game (http://www.espgame.org) Types of Tags System Design and Attributes • Identifying What (or Who) it is About • Tagging Rights • Identifying What it Is • Tagging Support • Identifying Who Owns It • Aggregation • Refining Categories • Type of Object • Identifying Qualities or Characteristics • Source of Material • Self Reference • Resource Connectivity • Task Organising • Social Connectivity 4
User Incentives Tagging/Folksonomies Pros • Future Retrieval • Easy to use • Contribution and Sharing • Intuitive • Attract Attention • Cheap way of getting annotations • Play and Competition • Gives new users quick and simple impression of content • Self Presentation • Can aid browsing and search • Opinion Expression Tagging/Folksonomies Cons Tagging Conclusions • Inaccurate and irrelevant tags • Provide keywords to describe objects • Lack of stemming • Collaborative tagging provides a community view of object • Freely chosen tags can result in • Number of pros and cons to using – Synonyms tagging – Homonyms • Number of design approaches can be – Polysemy taken • Number of incentives for users to tag 5
Tagging Case Studies Tagging Case Studies • A number of large studies of tagging YouTube Flickr Del.ico.us have taken place Rights Self Tagging Self Tagging Free for all • Focus on different aspects of tagging Support Blind tagging Blind tagging Blind tagging • We will focus on three analyses of different systems, the systems analysed Type of Videos Photographs Links are: Object Source Participants Participants Web resource – Del.icio.us – Flickr – YouTube Del.icio.us – Golder and Del.icio.us – Golder and Huberman Huberman • Provide an analysis of collaborative • Analyse two del.icio.us datasets tagging systems • Dataset 1 • They study the del.icio.us system for – 212 URL’s organising bookmarks – 19,422 bookmarks • Analyse the structure of collaborative • Dataset 2 tagging systems as well as their – 229 users dynamical aspects – 68,668 bookmarks 6
Del.icio.us – Golder and Del.icio.us – Golder and Huberman Huberman • Users have variety in use of tags, some • Appears that most tags are added for have many, some have few personal use • Tags vary in frequency of use • Never the less they are still useful to the public • However, stable patterns emerge in tags • They believe that consensus choices that emerge may be used on a large • Adding numbers of infrequently used scale to describe and organise how web tags/opinions, does not disrupt the documents interact with one another, general consensus and also can be used to make recommendations Flickr – Marlow et al Flickr – Marlow et al • Present a study of the photo sharing • Present a model for tagging systems and tagging system Flickr • Compare tagging on Flickr with that of • Looked at the dynamics of the Flickr del.icio.us and hope to expose interesting trends • Dynamics of interaction and and topics in the Flickr participation are different to Del.icio.us • Set of 25,000 users for individual • To be expected as they are different analysis models of tagging, and different user • Set of 2,500 users for network analysis incentives 7
YouTube – Halvey and Keane YouTube – Halvey and Keane • More interested in user goals when • Views of pages match distributions using and searching in a tag based found in web search system • Views from navigation match • Examine why some videos are viewed distributions found in web navigation more often than others • Found that in general the more tags that • Investigation of user interactions to see users provide the more likely that a if they vary in Web 2.0 video will be watched, up to a certain point • Analysed104,465 video pages from 57,639 users YouTube – Halvey and Keane Tagging Case Studies • In general users are consumers rather than • Briefly seen a number of analyses of creators of resources for the service tagging systems • Users of YouTube do not use the social tools • A number of different approaches were unless they gain a benefit investigated, from a number of different • Videos are recommended because they are aspects popular, not popular because they are • Although tagging and tag based recommended systems seem random, there are a • Videos receive the majority of their views in number of regularities the first couple of days 8
Tag Clouds Tag Clouds • Visual depiction of user generated tags • There are a number of uses for tag clouds, these include • Normally there is a weighting associated with the tags – Browsing – Search • Can be alphabetised – Impression Formation/Gisting • Importance of a tag can be represented – Recognition/Matching by font size or colour • Tags are usually hyperlinks that lead to further information Tag Clouds – Rivadeneira et Tag Clouds – Rivadeneira et al al • Investigated the use of tag clouds for • The use of different font sizes had an impression formation effect • Experiment 1 • Layout has an influence on the effectiveness of the users in performing – Influences of tag cloud attributes on low- tasks level cognitive processes • Experiment 2 • Some of the results can be attributed to westernised reading – Effect of font size and word layout on impression formation and memory 9
Recommend
More recommend
