XIST Inc.

Home > XIST Presentations > E-news Presentation
E-news Presentation

What's the word on e-news?

Presented by: John Savage, MLIS (masters of info sci), BBM (bachelor of biz management), CRA (appraisal designation)

Who am I?

Media monitoring librarian from 1996 to 2001 at Industry Canada

  • Proposed and co-developed Enews Expo in 1998.
  • Trained over 500 new users of enews systems
  • Redesigned the business process of Communications Branch clippings staff
  • Redesigned and recommended IC system changes

Media Monitoring Consultant

Worked for Ian Christie / Dom Idone of PWGSC to:

  • Help negotiate new EMM licence (saved $200,000)
  • Develop a business case to develop a shared enews system to save $$$
  • Worked with Transport Canada to redesign enews system:
    • Into a full feed
    • Expanded search logic
    • Monitoring of tv closed captioning
    • Develop search logic for Communications and various branches, including security.

What's the word with enews?...words

  1. How did e-news get here
  2. What's coming
  3. How keyword logic is crucial to good e-news design
  4. The role of the e-news Audit
  5. Redesigning your search logic to fit your needs
  6. Weaving words like fishing nets
  7. How to apply an e-news audit
  8. Future of e-news with new media
  9. XIST's services to support media monitoring

How did electronic news get here?

  • Electronic news was primarily available through online databases, such as Infomart and Dialog via libraries
  • News was physically clipped up until early to mid-1990s for media monitoring purposes?
  • Rise of commercial aggregators (Hoovers, Dow Jones)
  • Home grown aggregator systems with bulk content (NewsDesk, TC's NewsManager)

What's to come for enews systems?

  • Home grown monitoring of tv broadcast closed captioning (already happening)
  • Department wide shared systems?
  • Federal wide shared systems and shared clippings programs?
  • "New media" media monitoring (Blogs, web sites, email and phone tapping)
  • Focusing on content:
    • More content / news sources (e.g. under EMM licence)
    • Better search logic to control retrieval of more content

What is the next trend in Keywords and Search Logic?

  • Understanding that enews systems are only as good as the content (keywords, search logic) that go into them.
  • Enews systems are only as good as the searches.
  • Keyword content is King.

Fishing metaphor: How big are your nets?

  • All systems work like a fishing nets to catch the stories that suit their needs.
  • Big nets are used to capture news related to all the subjects of the department's wide mandate.
  • Small nets are used to capture news related to a narrower interest in subjects.

Problems

If the size of your net doesn't cover the breadth of your interests, then the stories don't get captured.

This results in:

  • Embarrassment
  • Stress
  • Missed opportunities.

How big are you're the holes in your nets?

Search logic and keywords determine how big the holes in the net will be.

  • Big holes may be designed to catch big fish or big stories.
  • Small holes may be designed to catch all sizes of fish or stories.

Problems

Sometimes the size of these holes isn't by design. Stories slip through the large holes that were supposed to be smaller.

This results in:

  • Embarrassment
  • Stress
  • Missed opportunities.

Nets that need mending

Every net needs a good mending from time to time. A search logic / keyword audit is the key.

How do you know you need mending?

  • Over-reliance on a few sources for terms (e.g. Minister's Office, subject experts)
  • Stories are missed that result in stress
  • Departmental mandate changes (new adminstration, new Ministerial interests)
  • New users desire news that suits their interests.

The Audit

A strategic look at your net / search logic.

What are the needs of your user groups?

  • Program personnel (researchers)
  • Minister's office
  • Clippings people
  • Executives
  • Librarians

The Audit

What are your subject interests (big vs small net)?

What types of stories do you want (big versus small)?

What are your business process?

  • Does retrieval fit with existing search logic?
  • Does system fit with needs of clippings staff?
  • What works, doesn't work, needs improvement?

Are there outdated terms (e.g. Information highway)
Are there new terms (e.g. ground rage, bio-terrorism)

Redesigning your net

Start with your subject focus to determine how big you want.

  • Solution may likely require improved vocabulary and search logic.

Talk with your clients to determine big versus small story interests.

  • Solution may likely require improved vocabulary and search logic.

Talk with your media monitoring staff to see their needs and stresses.

  • Solution may likely require redesigning interface and results display.

The role of search logic in the search

  • To capture relevant news
  • To limit the number of stories coming in
  • To save $$$ (e.g. problems with rogue users)
  • To save $$$ and improve control by switching from selected feed to full feed

Searching effectively

  • Goals of good enews retrieval
    • Recall
    • Precision
  • Tactics for good enews retrieval
    • Using a rich source of terminology that's relevant
    • Advanced search logic capabilities
    • Testing, testing, testing
  • Good enews business process
  • Good enews system design

First goal of good enews retrieval

RECALL:

The percentage of relevant articles retrieved as a ratio to the total number of relevant articles available.

  • This is important to the Minister's Office to know that all relevant articles have been retrieved. The Minister won't be surprised if there is high recall.

How to improve recall?

  • Ensure all relevant keywords are used.
  • Truncate terms, for instance automobile, automotive should be inputted as auto*
  • Keep search logic broad enough to encompass subject.
  • Test and revise search logic over and over until a good percentage of relevant articles are found in relation to unwanted stories.

How to ensure all relevant words are used to improve recall?

Use:

  • Survey subject experts
  • Thesauri / glossaries specific to your departments
  • Your department's web / intranet search logs
  • French-English dictionaries
  • Metadata (view source)
  • Termium Plus http://www.termiumplus.translationbureau.gc.ca/site/termium_plus_e.html
  • Search the news and Internet to confirm proper usage and to find new terms.
  • Outlines of departmental programs
  • Other Departments' media monitoring teams

Where to get help with new terms?

  • Departmental library as a source for: Departmental thesaurus, subject specific thesauri (e.g. medical thesauri), style guides used by journalists, and dictionaries.
  • Thesaurus manager in your department.
  • Internet and Intranet management for search logs that have keywords used by both public and internal users and metadata.
  • Access subject experts through various branches and your knowledge management team.
  • Termium Plus through the GoC's Translation Bureau.

Second goal of good enews retrieval

PRECISION:

The percentage of relevant articles to the total number of articles retrieved.

  • High precision means less effort to review search results. Less work means greater staff efficiency.

How to improve precision?

  • Use exact terms, phrases, names, dates.
  • Use Advanced search logic techniques to:
    • Combine more than one term into conditional relationships
    • Indicate where they must appear in the text (e.g. Title, Headline, Lead Paragraph, Date or Author field)

Risk of high precision:

High precision can reduce recall, if search logic is too precise. If searches are too precise, relevant stories will not be retrieved.

This is a major problem if precision is increased to reduce the number of stories retrieved and as a result, good stories are missed.

The ultimate goal!

The ultimate goal for any enews system should be:

High recall with high precision.

To ensure all relevant stories are accessed and that a high percentage are relevant.

How to improve both precision and recall?

Use words effectively:

  • Add more terms (improves recall).
  • Truncate terms, using roots, such as Auto* (improves recall).
  • Use exact phrases, names, titles (improves precision).

How to improve both precision and recall?

Use Advance Search Logic:

Boolean searching

  • AND (improves recall in relation to phrase searching and precision)
  • OR (improves recall)
  • NOT (improves precision)

How to improve both precision and recall?

Proximity and field search operators

  • Headline and Lead Paragraph [HLP] (improves precision)
  • NEAR3 (improves recall and precision versus phrase searching)
  • SENTENCE (improves recall and precision versus phrase searching)
  • ETC.

How to improve both precision and recall?

Nesting

  • Use of parenthesis to link concepts (improves both recall and precision)
    • eg. (concept) and (other concept)
      ((car or auto or suv or truck) near10 (pollution or emissions)) OR ((boat or ship or tanker) near10 (ballast or effluent or wastewater or bilge))

Group Concepts (Facets)

To improve recall and precision:

  • Break large searches into subjects, also called concepts or facets.
  • Develop search logic for each topic using rich source of terms
  • Develop equivalent searches in French using literary warrant where possible for equivalent terms (web sites, TC Thesaurus, TC metadata, news)

Test, Test, Test, Test

Testing is absolutely necessary and crucial to effective search results.

Test for:
Good recall and good precision by comparing ratios of relevant articles from broad searches to narrow searches. This is done by using the same sample.

Good eNews Management (Governance)

  • Develop an eNews Glossary / Thesaurus for terms in both English and French to call upon for new searches.
    • Track sources
    • Possibly involve library if they can support your efforts.
  • Develop searches based on subject experts interests.
  • Review and edit searches on a daily, weekly, monthly basis.

Good eNews System Design

  • Involve subject experts on a routine basis to develop new searches and delete old ones.
  • Break large searches into topic searches.
  • Have a view for editor to see all topic searches as a group to also help news clippers.
  • Consider international sources to see what is being said about department and topics of interest abroad (e.g. Factiva)

Search Logic Audit

Conduct a search logic audit periodically:

  • For small nets [e.g. Communications purposes] at least semi-annually
  • For large nets [e.g. whole departments like Industry Canada] daily to quarterly.

Why?

  • News and needs change and so do terms. [e.g. my work Sept. 11, 2001 for Transport Canada]

The Audit Again…

A Search Logic Audit should contain:

  • A review of user needs and levels of satisfaction
  • A review of existing search logic
  • A review of business processes and workflow
  • Recommendations for new possible terms and search logic development, if necessary.

Future of enews?

New media:

  • Web Blogs (web logs that are being run now both by amateurs and professionals)
  • Web site content tracking (tracking think tanks publications, Opposition Parties' web sites, other government web sites, etc.)

The role of words/logic with new media?

The same as with enews. Words and search logic are the fuel of any system. A great system is only so good as those who develop the search logic.

Example: Broadcast tv closed captioning monitoring

XIST's media monitoring services

  • Audits
  • System design for electronic print and broadcast tv closed caption content
  • Search logic development
  • Content licensing negotiation
  • Client training
  • Business process design

John Savage

john.savage@xist.com

Suite 402, 176 Gloucester Street
OTTAWA ON K2P 0A6

613.234.9621, ext. 238 / 1.888.ASK.XIST
FAX 613.234.9564

www.xist.com



XIST logo
© XIST Inc. 2009 
Phone: (613) 234-9621  1-888-ASK XIST   Fax: (613) 234-9564   Email: xist@xist.com
Office Location:    176 GLOUCESTER ST., SUITE 402    OTTAWA ONTARIO    K2P 0A6
Mailing Location:    P.O. BOX 4938 STATION "E"    OTTAWA ONTARIO    K1S 5J1