EnterpriseSearchCenter.com Home
  News   Features   White Papers   Research Reports   Web Events   Conferences  
 
RESOURCES FOR EVALUATING ENTERPRISE SEARCH TECHNOLOGIES
July 12, 2006

Table of Contents

ESCenter Premieres Exclusive Articles Series
Featured Content: Principles of Effective Search
This Week in the Demo Center
E-mail search with confidence
Enterprise questions, enterprise answers
Findology Interactive Media Launches Content Map
StoredIQ Offers Integration with EMC Centera
LexisNexis Introduces Intelligence Analysis Solutions
Categorize completely
veotag Launches Service for Online Authors

ESCenter Premieres Exclusive Articles Series

The Enterprise Search Center has invited a number of experts to write exclusive articles for us to share with you. We'll be publishing an article every couple of weeks, starting with a piece today called "Principles of Effective Search," by James Robertson.

JAMES ROBERTSON is the managing director of Step Two Designs (www.steptwo.com.au), a leading vendor-neutral intranet and content management consultancy located in Australia.

In PRINCIPLES OF EFFECTIVE SEARCH, Robertson shares highlights from a study he conducted on improving intranet search results.

Download the Free PDF.

Back to Contents...

Featured Content: Principles of Effective Search

 There are a number of fundamental principles that underpin the design and delivery of successful search. These relate to the way that staff use search, and the types of tasks they are trying to complete. They are also built on an understanding of staff motivations and behaviors. These principles are not new or radical, rather they are built upon the observation of many users when they search. They distil down common behaviors and approaches that can be analyzed to help design better and more effective search tools.

These principles apply across all forms of enterprise search, regardless of whether users are searching an intranet or the corporate website or document management system. While specific user tasks may vary, the same fundamental principles define an effective search solution. It is useful to encourage project teams to return to these principles when making design decisions, as they will often provide insight into the right approach to take when confronted with several competing options.

At a Glance: Key Principles
These are the fundamental principles that underpin an effective search engine:

  • Search should work like magic, so that users can type in a few words, hit the search button, and find the information they are looking for. • Search results should be relevant, with the most useful pages or documents in the first few hits. Staff must also be able to assess the relevance of the search results in order to trust it.
  • There should be a strong information scent that allows users to quickly and easily distinguish between search results, helping them find the required information.
  • To inspire staff to actually use the search tool, they must trust it. Users must be confident that they will be able to find the desired information, and that using search will be quicker and easier than any other mechanism.
  • Search must be designed with a strong understanding of search needs and behavior of staff throughout the organization. This is particularly important when delivering search for specialist users.
  • Work should be done on behalf of search users in order to make it quicker and easier to find information. This delivers real (and measurable) productivity benefits to the organization.
  • At the end of the day, usability is not a training problem. As it is not practical to train all staff to "search better," the search engine needs to be designed to match current skill levels and experience.

These principles are explored further in the following sections, with examples of what they mean in practice, and implementation tips.

Works Like Magic
Thanks to Google, users expect to be able to type in a word (or two) and find the page they are looking for, preferably in the first few results. This is not an unreasonable expectation. At the most fundamental level, search is supposed to make it quick and easy for staff to find things, thereby saving them time and improving productivity. This can be distilled down to a very simple concept: search should work like magic. As much as is possible, search should always give staff the information they need, somewhere in the first few results.

Staff should not have to learn complex search functionality, or spend time carefully considering the most effective search terms and options. Regardless of what the user is searching on or which system the user is searching, the right results should be returned every time.

Delivering this standard of search experience involves making under-the-hood improvements to the search engine to improve its effectiveness. This includes tweaking weightings and other configuration settings, as well as exploring the use of spell-checking and other enhancements.

Other practical steps include the following: Implementing search engine synonyms, which match up equivalent terms (such as bike and bicycle). This addresses the mismatch between what a user is searching for and the contents (and metadata) of page and documents. They can also be used to resolve spelling mistakes and other typos. Implement search engine best bets, which are a hand-created list of key pages for common searches. For example, if a user searches for leave, they are almost always looking for either the leave form or the leave policy.

Poor search results actually provide information that can be used to refine future user experiences. Thus, it is important to monitor failed searches and take corrective activity to tune results accordingly. This should be done as part of the regular monthly schedule of search engine maintenance and enhancement.

Works Like Magic
Thanks to Google, users expect to be able to type in a word (or two) and find the page they are looking for, preferably in the first few results. This is not an unreasonable expectation. At the most fundamental level, search is supposed to make it quick and easy for staff to find things, thereby saving them time and improving productivity. This can be distilled down to a very simple concept: search should work like magic. As much as is possible, search should always give staff the information they need, somewhere in the first few results.

Staff should not have to learn complex search functionality, or spend time carefully considering the most effective search terms and options. Regardless of what the user is searching on or which system the user is searching, the right results should be returned every time.

Delivering this standard of search experience involves making under-the-hood improvements to the search engine to improve its effectiveness. This includes tweaking weightings and other configuration settings, as well as exploring the use of spell-checking and other enhancements.

Other practical steps include the following: Implementing search engine synonyms, which match up equivalent terms (such as bike and bicycle). This addresses the mismatch between what a user is searching for and the contents (and metadata) of page and documents. They can also be used to resolve spelling mistakes and other typos. Implement search engine best bets, which are a hand-created list of key pages for common searches. For example, if a user searches for leave, they are almost always looking for either the leave form or the leave policy.

Poor search results actually provide information that can be used to refine future user experiences. Thus, it is important to monitor failed searches and take corrective activity to tune results accordingly. This should be done as part of the regular monthly schedule of search engine maintenance and enhancement.

Ensure Relevance
It may seem to be a statement of the obvious that search results should be relevant and useful. Unfortunately, it may be obvious in theory, but not evident in practice. As much as possible, this means that the information that the user is looking for should be listed at the front of the results.

It has often been shown that users rarely look beyond the first page of results, and many never examine more than the first half-dozen hits. This means that while the desired page will almost certainly be included in the search results, there is no value in having it as hit number 140. If the desired page isn't listed in the first 20 hits, it will generally never be found, and the search engine will be deemed a failure.

The fundamental purpose of the search results page is to help staff find the site page, information, or document they are looking for. The results should therefore be designed to make it easy for staff to distinguish between the hits, and to quickly identify the desired information.

Users do not read search results pages in detail. Rather, they quickly scan the pages, looking through the search results. In practice, they are applying a pattern matching approach, looking for the first page that appears to match what they are looking for.

In order to best match these usage patterns, search results should have a strong information scent. No, we haven't caught a whiff of any actual smell-related search technologies. The scent trail here is actually a more visual one: results should provide a clear indication of what is being presented, displayed in a way that allows users to quickly and easily distinguish between hits.

Items with a strong information scent provide a clear context for each result, allowing users to assess the relevance of each hit. Equally importantly, they don't overload the user with too much (and often irrelevant) information that makes it hard to pick out the most useful detail.

The requirement to provide a strong information scent therefore demands that the search results be simplified, down to the key details that allow users to quickly scan the results and locate the desired page. There are a few practical steps you can take to provide strong information scent:

Start by stripping out as much as 50-80% of the information and functionality on the default search results page provided by the search vendor. While the vendor is trying to show off the full functionality of the product, this level of complexity can be overwhelming for general search users. Eliminate over-complex or under-used features, such as re-ordering search results, "find similar" functionality, or excessive detail for individual search hits. While this may be relevant for specialist users, the complexity impacts the usability for general search users. Take time in developing a clean and attractive layout for search results that increases both readability and scanability.

Trustworthy Search
Users will only use search if they have confidence that it will be useful for them. This trust is built up (or lost!) over time, based on the direct experiences of using the search engine. If a staff member is repeatedly unable to find information using the search, or they find it very frustrating or difficult to use, they are unlikely to come back.

This is a crucial consideration when designing a search solution, for no amount of extra features will have an impact if users aren't actually using the search. In practice, staff trust in the intranet search means a number of different things:
  • Trust that required information will quickly be found.
  • Trust that the pages returned are useful and relevant.
  • Trust that all key information is being searched, and that nothing important has been missed from the search results.
  • Trust that the search is easy to use and well understood.
  • Trust that using the search is quicker and easier than using other mechanisms (such as asking the person in the next cubicle).

These aspects cannot be directly assessed by users of the search engine. Instead, they rely on their accumulated experiences when making a value judgment about whether they trust the search.

Practical steps for building trust in the search engine include the following: Make it easy for users to assess the relevance of the results, by focusing on the details that give strong information scent, while eliminating unnecessary clutter. Present a clean and professional image for search that conveys the sense that time and effort has been spent on delivering an effective search solution. Monitor common searches, and run these to assess the quality and relevance of the results. Where required, take steps to improve these searches, such as by adding metadata or synonyms, or by further tuning the search engine configuration.

Matching Search Behavior
Search must be designed with a strong understanding of search needs and behavior of staff throughout the organization. There is no single best way to design search. Instead, a given search engine should match the unique needs and issues that will be encountered in each organization. This is particularly the case when search will be used by specialist staff such as researchers, librarians, or lawyers. These staff will have quite different search behaviors than general search users.

The design and implementation of the search engine should therefore be guided by the results of user research conducted with actual staff in the organization. This includes using a range of techniques such as interviews, workplace observation, and contextual inquiry. This can be supplemented by surveys and focus groups (although these are not recommended as primary research techniques).

This research may highlight that the needs of specific groups of users varies greatly from general search users. For example, the screenshots below show a search interface specifically designed for corporate lawyers searching through a repository of past legal judgments. While this is on the surface a complex interface, it is perfectly suited to the specific needs of this user group. By working closely with the actual users, a tailored search interface can be created that delivers real productivity benefits.


 

Work for the Workers
At the end of the day, the role of enterprise systems such as the corporate intranet is to make life easier for staff, by helping to more quickly complete tasks or find information. If the intranet is not providing these benefits, then it has failed in its primary goal.

With respect to search, this means that designers should look do the hard work on behalf of staff, so that they find searching easy and satisfying. It comes down to mathematics: The benefits delivered by improving the effectiveness of search are multiplied by the number of staff in the organization. Saving each staff member 5 minutes quickly adds up to a large sum in employee ROI.

In this way, the behind-the-scenes improvements to search can significantly improve the productivity (and satisfaction) of staff throughout the organization. In practice, even a few days devoted to improving the design and configuration of search can have a dramatic impact on the effectiveness of the search solution.

Train the Search, Not the Searcher
While it may be possible to provide training for small groups of specialist users, it is not practical or realistic to train every staff member within an organization on key searching skills. While it would be wonderful if all staff searched well, the reality is that search tools need to be designed to match the existing skill levels and experiences of staff.

This means designing search to match the "lowest common denominator" within the organization, so that staff can successfully use search, even without any real understanding of how search works. For better or worse, people today have a Google-ized expectation of search: they want to enter a few terms that make sense to them and voila, have relevant, usable search results appear . . . which generally comes back to ensuring that the search works like magic.

Getting Search Results
There is much that can be done to improve the effectiveness of search. Regardless of the power of the search engine, simply installing even the most powerful tool "out of the box" and walking away is not enough.

Instead, design work is required to deliver a search solution that works well for any given staff so that it delivers the desired productivity benefits. When conducting this design work, project teams should return to the fundamental principles outlined in this article. These principles will help to focus efforts, and will target efforts to the areas that will have the greatest impact on staff. They can then help to guide the exploration of more powerful features, such as the use of taxonomies, faceted searching, federated search, and more.


This article was adapted from the "Improving Intranet Search" report written by James Robertson of Step Two Designs. This 115 page report provides many practical suggestions and examples for improving the effectiveness of search, and can be purchased and downloaded from the Step Two Designs website (www.steptwo.com.au).

Back to Contents...

This Week in the Demo Center

This week Siderean has added a flash demo illustrating its Seamark Navigator 4.0 product. Visit our free demo center to take it for a spin. While you're there, you can also check out the demos from: Coveo, exalead, ISYS, MondoSoft, NorthernLight, Synomia, Thunderstone and Vivisimo.

Back to Contents...

E-mail search with confidence

Fortiva has introduced what it says is the first search performance guarantee for archived e-mail, ensuring instant search results regardless of the amount of data stored.

The company explains the guarantee is made possible by its grid storage and distributed search architecture, which allows users to search through archived data with the speed and ease of an online search engine. In addition to ensuring efficient legal and regulatory discovery, this capability allows Fortiva users to improve productivity by providing search functionality directly from Outlook, says Fortiva.

With the new search performance guarantee, Fortiva promises that 95% of all customer search requests will occur within five seconds or less. In the event that the customer experiences slow search performance, Fortiva will credit the customer a pro-rated portion of his or her annual billing amount.

Back to Contents...

Enterprise questions, enterprise answers

Teragram has released Direct Answers for the Enterprise, an offering designed to search documents within corporations and provide users with simple answers. The first natural language processing (NLP) technology of its kind at work in the corporate setting, Teragram claims, Direct Answers for the Enterprise scans structured databases such as Excel spreadsheets and sales lead databases, as well as semi- and unstructured documents like text-based reports and Web pages, to provide comprehensive answers derived from multiple sources of information. Direct Answers integrates seamlessly with existing search software and does not require hardware installation. Additional benefits include:

  • multilingual query processing--parses user questions into meaningful search terms;
  • federation--aggregates search results from multiple sources controlled by the enterprise (including other search engines);
  • scalability--accommodates unlimited amounts of information, depending on server capacity; and
  • speed--scans up to 50 queries per second.

Teragram explains that Direct Answers for the Enterprise works by parsing a user's initial query and analyzing it syntactically and semantically in a process that takes milliseconds. By processing content in document repositories, enterprise content management systems and specialized enterprise applications such as e-mail programs, the product is able to deliver answers in seconds, the company says.

Back to Contents...

Findology Interactive Media Launches Content Map

Findology Interactive Media, Inc., a provider of customizable online marketing products and services, has launched a new technology for website publishers, Content Map. The service is designed to enable publishers to increase page views and maximize their site's "stickiness" and advertising real estate, by allowing visitors to find and access site content related to their expressed interests.

Content Map allows publishers to utilize Findology's contextual search and content technology. The service inserts hyperlinks into web page text that visitors can click on and view a list of related content within the same site. Additionally, Content Map analyzes a web site's content and matches sponsor listings to each of its pages. The sponsor listings are customizable to fit the look of any page, enabling publishers to monetize their site with premium advertisements, while maintaining their site's branding and design.

The new offering from Findology represents the company's efforts to provide publishers with the customized tools needed to maximize site revenue through targeted advertising inventory and innovative search technology. Content Map is now available to new and existing Findology partners.

(www.findology.com)

Back to Contents...

StoredIQ Offers Integration with EMC Centera

StoredIQ, Inc. a provider of automated Information Classification and Management (ICM) software, has announced it has completed integration between the company's ICM 5000 content-based classification and policy management software and EMC Centera CAS. Centera is a software-driven storage platform that provides capabilities for the business value from unchanging or infrequently changing digital assets. StoredIQ is also now an EMC Velocity2 ISV CAS Specialty Advantage partner and will work closely to market and sell the combined solution.

The integration combines StoredIQ ICM 5000, a content-based classification and policy management solution with, EMC Centera, an archive storage platform, giving organizations the ability to discern what information is stored in Centera, based on the content of that information. The relationship allows for contextual classification and automated policies that can copy, move, set retention, or apply litigation hold onto information going into Centera. The appliance-based software also comes in a four and eight-node cluster offering.

(www.storediq.com; www.emc.com)

Back to Contents...

LexisNexis Introduces Intelligence Analysis Solutions

LexisNexis U.S., a provider of information solutions and services, has announced the launch of its Intelligence Analysis Solutions to enable analysts in the intelligence community to pinpoint and identify relevant information contained within volumes of disparate data.

Key applications of Intelligence Analysis Solutions include: Open Source Analysis to enable analysts to pinpoint specific threat information contained within disparate, multi-sourced data sets such as news articles, transcripts, scientific journals, public records, and web-based content; Maritime Domain Awareness to allow analysts to identify links between cargo, containers, businesses, crews, and vessels to enable risk assessment for interdiction and mitigation of threats before they reach U.S. shores; and All-source Analysis to enable intelligence analysts to enrich classified data sets with open sources of information for enhanced analysis and contextual insight to detect and anticipate threats to national security. LexisNexis Intelligence Analysis Solutions can be delivered via secure access through the web or deployed on-premise within a trusted enclave for use in an agency's secure and classified environment by LexisNexis Special Services Inc., a company established to deliver LexisNexis solutions to customers with classified missions.

LexisNexis U.S. has also launched LexisNexis Total Practice Advantage, a new solution that brings together the software and web tools small firm attorneys and their staffs can use to more effectively manage their business and grow their practice. A combination of practice management, research, client development, investigation tools, and more, Total Practice Advantage is software tightly linked to exclusive online LexisNexis content. The company's practice management software, Time Matters, is the backbone of this offering, which includes content, capability, insight, and investigative tools from LexisNexis brands including Shepard's, Martindale-Hubbell, CourtLink, HotDocs, Mealey's, Matthew Bender, Michie, Collier, SmartLinx, and LexisNexis Company Dossier. LexisNexis is offering a general litigation version and state-specific litigation versions for Florida, Georgia, Pennsylvania, Virginia, Connecticut, California, Indiana, Massachusetts, Maryland, New York, Ohio, Tennessee, and Texas.

The company has also announced the launch of LexisNexis Advanced Government Solutions to enable customers across federal, state, and local government to transform data into mission-critical decisions. As part of the new offering, LexisNexis will deliver three distinct solution sets: Intelligence Analysis Solutions enable intelligence analysts to anticipate and detect threats to national security by pinpointing critical information contained within volumes of data; Investigative Solutions enable investigators to locate suspects, find missing children, and solve cases by providing access to critical information; and Screening and Identity Verification Solutions enable government agencies to secure and protect access to critical systems and facilities and mitigate the problems associated with entitlements, fraud, and improper payments.

(www.lexisnexis.com; www.reedelsevier.com)  

Back to Contents...

Categorize completely

Inxight Software has announced Categorizer 5.0, a new, hybrid categorization engine with automated techniques and editorial controls to ensure accuracy. Version 5.0 can also deploy third-party and industry-standard taxonomies and includes a "workbench" feature designed to facilitate the creation of testing of taxonomies.

Categorizer 5.0 is available as a component within the Inxight SmartDiscovery Extraction Server platform and works in concert with the company's ThingFinder extraction and Summarizer modules to structure unstructured text in more than 30 languages. Using the new workbench capabilities, categories can be defined manually (rules only), with learn-by-example only, or using both techniques together.

Further, Inxight says, the workbench provides a robust set of testing tools for taxonomy editors to define, maintain and test their taxonomies before deployment on a SmartDiscovery server. It also includes many other new features, including an interface to assign training documents, a rule-writing helper and a testing environment for the creation of truth sets and accuracy reporting.

Back to Contents...

veotag Launches Service for Online Authors

veotag, a provider of resources and tools designed to help solve problems associated with audio and video on the web, has launched a service that allows online video "authors" to add clickable, explanatory text to almost any online video to improve audiences' viewing experiences. The service, available for free to individuals, is designed to enable video authors to describe what's inside a video, and where, and allow viewers to go to specific, desired parts of the video by clicking on related text. veotag also enhances audio files.

veotag enables individuals to "annotate" with clickable text just about any video (or audio file), and it requires no special software or technical skills. After a video (or an audio file) is "veotagged", viewers can navigate the video as if they're skimming a book with a good table of contents. There is no charge for veotag's basic consumer service. To "veotag" a video, the author goes to veotag's website and enters the URL of the video (or audio file) he or she wants to annotate. veotag locates the video, and then begins playing it on the author's screen. Whenever the author wants to add a comment, he or she can stop the video and type desired text in an adjacent text box, and then resume playing the video. When the author has finished adding all the text, he or she saves the text to a personal veotag account. veotag then gives the author a veotag.com URL to share with anyone--or everyone.

Viewers of the "veotagged" video can click on the text created by the author and go right to the corresponding scene. Viewers reach the veotagged video by clicking on the veotag.com URL associated with it. They no longer need to know the URL of the original video. In addition to making it go directly to specific parts of a video, veotagging allows everyone on the web to find and use video and audio: veotag gives its authors the option of submitting their veotagged videos and audio files--clickable text and veotag URL--to the search engines.

(www.veotag.com)

Back to Contents...
 
[Newsletters] [Home]

Problems with this site? Please contact the webmaster. | About ITI | Privacy Policy