We'll talk about...

• How are keywords used?
• What are your challenges?
• Types of terminology tools:
  • Generating keywords and tags
  • Extracting concepts and identifying language patterns
  • Finding alternative words and phrases
  • Borrowing existing vocabularies
  • Browsing subject domains

Generating keywords and tags

Tags, clouds, divining patterns

• Short words - or concatenated phrases - for "aboutness"
• Familiar to many people now, so can be useful with SMEs
• Clouds add a patterning dimension for visual interpretation

Generating keywords and tags

Cloud Generators (online)

• Online - just feed a URL or some text
• Usually single terms, not phrases
• SEO's Keyword Density Checker is a little different:
  • shows frequency of terms in a list as well as a cloud
  • direct feedback on heavily-used terms and little-used terms
  • their interest will prompt other tools to be developed

Generating keywords and tags

Cloud Generators (embedded)

• Focused on embedding a cloud into your web page, rather than on term identification
• Yes, they're "cool" - but useful to us?
• Try adding to a development version of the site, for non-public analysis
• Text size variance offers hints for frequency of use... but not necessarily relevance

Extracting concepts and patterns

What is concept extraction?

• Automated processing of text to identify word/phrase patterns
• Can consist of a series of automated steps, such as:
  • shredding text into words and phrases
  • proximity and frequency analysis
  • identifying parts of speech
  • semantic analysis of phrases
  • word sense and disambiguation
• Not easily used as an end in itself... part of the process
• Close, but not the same, as "entity extraction"

Extracting concepts and patterns

Term Extractor

• Part of a European research project at an Italian University
• Not many of these out in the "free web" space
• Simple approach to uploading one or more files for rapid analysis
• Quite a lot of control over analysis... if you can understand what the controls offer!
• No rules engine for further tuning/processing against a specific set of content

Extracting concepts and patterns

Yahoo Term Extractor API

• If you have a developer handy... this can be very useful!
• Available as a service to be called by other applications
• Many possibilities for identifying terms and relationships
• No rules engine for further tuning/processing against a specific set of content

Extracting concepts and patterns

FeatureLens Text Pattern Analyzer

• Not currently available, but something for you to think about
• The primary focus is not extraction, but analysis... just what we need!
• Useful for identifying patterns and relationships across texts or within complex texts
• Clear, understandable interface, as well

Finding alternative words and phrases

Recognition is easier than recall, even in formative stages

• You are not your user... so your language may not be theirs...
• Sites are available to help look at language alternatives

Finding alternative words and phrases

Thesauri

• Quick, simple way to look up alternatives and stay out of ruts
• A range of interactions and depth of information to choose from

Finding alternative words and phrases

WordNet (Princeton University)

• Extensive, available terminology database
• Word usage and alternatives
• Used in many research projects, particularly for natural language interpretation
• Multi-lingual

Borrowing existing vocabularies

A wealth of terminology helps broaden local vocabulary

• Apologies for one error in paper -- OpenDirectory Project is at: www.dmoz.org
  
  
• Isn't the Web as a collective sharing space wonderful?
• Just because the vocabulary sets are available doesn't mean it's effort-free
• Many different formats, structures, purposes -- and licenses
• Some examples...

Browsing subject domains

People are familiar with their subject domains

• People come to sites with a context -- their experience in the wider world
• Reality check and align your own vocabulary with sites that are probably familiar to your users
• Cross-checking with known subject domains also helps reinforce discussions with SMEs, if consensus becomes a problem
• Cloudalicio.us is different: it shows you trends in keyword usage, which may tell you things about the users' awareness and receptivity to certain terms
• Some examples...