Glossary of Search Terms

Search Terms

CORE SEARCH FUNCTIONS

The following tools are relatively core to all the providers surveyed, in addition The University is currently in a position to start making use of these tools, even though to get the best out of them does require active management of content.

Auto Correct

Auto Correct does exactly what it says, this offers users the option to search for misspelt words entered into the search engine.  This can be done automatically based on spellcheckers or using synonyms (see below)

Synonyms

Synonyms are similar to Auto Correct apart from the need to create these manually, synonyms allow a web master to add what they think the user is searching for based on local knowledge or research into search engine results.  For example a synonym could be added for "joining", so that when people search for registration they are also presented with results for "matriculation" or "Academic Registry" as both those terms are University specific.

Auto Population

Auto Population allows the system to attempt to predict what the user is typing based on content that has been indexed by the search engine (for example typing in "Craig Mid", the system would start to return "Craig Midddlemass", "Craig Middlemass - Prototypes", etc).

Weighting/Boost and Bury

Weighting is a common feature where using metadata (per page) or folder structures (hierarchy) or site maps the priority of content can be determined. For example rating an Edinburgh First page above a Timetabling Room Booking page which both contain information on the use of the Quad in Old College, would results in the Edinburgh First page being displayed above Timetabling page.

This feature can be static in terms of weighting assigned, but in some cases can be augmented by Boost so that a page can become the most prominent or bury so that it becomes the least prominent.

Spotlighting

Spotlighting allows the search engine to return a specific result page (similar to sponsors links on Google and Bing) where these will appear in a predominant way above the actual search results. These are keyed to only display when searches entered cross the topic that the spotlighted item relates to.

Faceting and Filtering

Both of these are used to segregate content so that search results can either be displayed through different section of the search results or by drilling down through categories of search results to discover the most relevant content without having to search for another term to refine the results.

Context Sensitive Searches

This allows for the search engine to behave differently depending on where the search was initiated from, this could range from altering the facets to display searches related to the site that you search from in conjunction to the whole University Website search or completely changing the search engine to be tune for the area. For example a search initiated from the PURE pages could be tuned to return results only from PURE with facets and filtering that reflect the content in the PURE database.

Preview Results Pane

Using auto completion and synonyms or just the bare search engine it is possible to create a preview pane that pops up with results as the user types.  This can be customised to present simple results, faceting or clustering.

OTHER SEARCH TOOLS

The following search tools provide a variety of functions but given the University's current level of sophistication in it's approach to search it is recommended that the be discounted.

Recommendations

Similar to synonyms in terms of management, recommendation can be used by some search sites to point users to content that they may be interested in (e.g. searching for an author of some teaching literature could also recommend other similar literature or an alternative author in that field of study).

File Type Display

Search engines can be modified to handle different file types, so that document or images can be displayed, listed or filtered to allow users to find particular files types more readily (e.g. a student searching for an image of a known researcher may want an image of what they look like rather than all the literature they have published).

Folder Filtering

Folder filtering presents users with a set of results the ability to drill down or filter on a specific area of results based on the folder that the content belongs to.

Geographical Refinement

Geographical Refinement allows results to be shown in relation to their location, this is more focused towards content that relates to a physical thing.  This way the user can focus on content that relates to objects located in a particular campus or building within the University (similar to how campus maps operates but looking at search results on the map).

Geographical Location

Where the user is willing to share their location, search engines can filter or refine the results to show results with a certain distance of their location.  Similar to Geographical Refinement but allows the users to act as a centre point for the results.

Graphical Refinement

Graphical Refinement is another form of filtering that displays the results in a table or graph that again allows the user to target search results, graphical refinement generally consist of a parameter such as date, data source, folder and display how often the search item appears in that area.  This could be useful where a user wants to explore more about their search by looking a group of results from a particular department (for example search for Economics will return searches from across the University but a concentration of results will be in the School of Economics, clicking into the School of Economic would then display all relevant results from that school).

Ratings Refinement

Ratings Refinement allows the user to filter their search results based on ratings assigned to content to other users or set by a web master.  This allows users to focus on how useful other users have found the content rather than just searching solely on relevance.

Cluster Refinement

Cluster Refinement is another method of filtering results which is based on the number of results relating to a cluster that the web master has defined.  These can reflect anything that the University wishes but does involve tagging content, the clusters can be multi-tiered in order for drilling down through results (for example a clusters could be for Research Publications then Publications by year).

Social Tools

Social Tools can be integrated into the sites to allow users to rate, comment and tag content without the need for a web master to do all of the ground work.  These allow data to be built up over time and then used within the search engine to provide filtering like the Ratings Filtering and Clustering searches.

Sentiment Analysis

Sentiment Analysis is offerred by a few providers where the search engine can trawl social sites or forums to look for positive, neutral and negative comments on a particular content or subjects.  The case study presented by one of the suppliers allows their search engine to trawl a University forum based on capturing feedback on teaching and then pulled the number of comments and assigns them a sentiment rating based on the number of positive, negative and neutral comments.  This has allows that institution to target areas for improvement based on real time feedback rather than waiting until the National Student Survey results are in to see where they are failing.

Structured Data

Structure Data generally refers to content that is designed for the web, the basis of the content being structured is that the content follows a format that allows metadata to be sourced from it and other attributes are always in the same place making a search engines job simpler as it can compare the structure content to other structured content to provide search results.

Unstructured Data

Unstructured Data generally relates to content that is not designed for the web, this can range from databases to documents. When a search engine interacts with these types of content it must search through the full content in order to classify it.  A lot of search providers offer searches of unstructured data the search engine can be connected to a database with no knowledge or context of what the database holds or represents, the search engine then derives what the content of the database is and factors this into its search results.   In all instances the search engine would need to be configured with permissions as to not crawl information that shouldn't be publically available.

Dashboards / Cockpits

Dashboard and Cockpits are a prime example of how search engines can push data to people rather than the traditional search for an item (pull method).  Dashboard and Cockpits can be customised to the user that is logged in either through the user's attributes (for example a staff member in Physics could be presented with content and search analysis related to Physics teaching and research without having to carry out any searches).  Dashboard and Cockpits require the user to be logged in but do allow for targeted searches.

Widgets

While the Dashboard and Cockpits provide a completely customised home page it is possible to take elements from them for use in the generic web search.  A number of providers allow the search engine to be reactive based on what the user is searching for or tailoring the search results based on the user being logged in and allowing the search engine access to attributes of the user.

 Matrix of offering