Enterprise search: search results browsing and navigation
Previous articles in this series have discussed enterprise search innovations in the areas of automated query improvement and relevancy ranking. The third key area of innovation is in the presentation of search results.
Where a good (that is, detailed and specific) query has been used, together with sophisticated
relevancy ranking, it can be expected that the desired documents are somewhere near the top of
the results list, within easy reach of the user. Where the data set is relatively small and the user
community have similar requirements, all that is needed in terms of results presentation is a format
that makes it fast and easy to identify the right documents.
As with other types of software, such as word processors, conventions have emerged for how search
results should be structured. The most basic form of this has:
- A clickable title line
- A short summary of the document
- Optionally, other information about the document derived from metadata
Some of the hard work needed to make this happen is done at index time. For example, in most organisations, Word documents typically do not have metadata such as title text. Instead, the search engine will automatically extract text from the body of the document to populate this field, typically taking the first line found.
Summaries are usually derived dynamically at search time, centring around words in the search clue, for example, the two sentences containing the highest density of query terms are formed into a summary.
In enterprise search applications, it can also be useful to display other metadata within search results, such as department, creation date or document format (often using an icon).
The key objective is to speed the browsing process, helping the user to quickly dismiss uninteresting documents and focus in on required information.
In some applications, it can be useful to display conceptual information such as “entity vectors”. For example, an additional line in each search result might display a list of the countries or companies mentioned in the article. The objective is to reduce the incidence of opening uninteresting documents, ideally to zero.
Dealing with Data Growth & Subjectivity
In enterprise search systems, two factors complicate matters. User satisfaction with enterprise search remains surprisingly low (this will be discussed in a later article in this series), and the most influential factors causing this are data growth and subjectivity. The innovation that most effectively addresses both of these challenges is “search navigators”, also known as dynamic navigation or facetted search. These are the hyperlinks, usually with little numbers after them, which enable users to easily drill-down into search results lists. Navigators are driven by metadata, and by far the most important aspect of implementing them is the capture and/or creation of accurate metadata, which is done at index-time.
Data growth means ever larger results lists for users to browse. Navigators provide a quick method to cut results lists back down to a manageable size.
Navigators are also the most effective way to deal with subjectivity. Different people may use the same simple search clue for entirely different reasons. Navigators enable the results set to be dynamically customised. In effect, they are another way of creating a better query by adding additional conditions to it. In this respect, nothing has changed since the early days of search engines. Early searchers would iteratively build a complex query, until the results set was cut down to a browsable size. Search navigators do exactly the same thing.
More detailed discussion of this subject can be found in this downloadable white paper, Making The Most of Search Navigators.
This concludes our review of the important innovations in enterprise search. The next article will explore enterprise search benefits and ROI.