9 2022 Defining ‘good practice’ in enterprise search

The first published bibliography of research into information retrieval was published in 1964 (Snoddy 1964) and covered the period from 1957–1961. Fast forward to 2021 when the IR Anthology was established with a database of over 40,000 papers (Potthast 2021). However, this collection is not fully comprehensive in its scope, and in total it could be that there are approaching 100,000 research papers.

In the case of academic research into enterprise search there has been only one research paper published which considers the way in which enterprise search is used across a single organisation. This paper (Lykke 2021) provides a wealth of data on how employees make use of enterprise search. The organisation was a Danish biotech company with 7500 employees. With any individual case study the issue is always the extent to which the outcomes scale to other organisations. Without going into analytic detail it is reasonable to assume that it does scale, certainly to other medium to large-scale high-technology organisations.

There have been a number of research papers which document the outcomes of projects to assess the way in which specific groups of employees (such as engineers) make use of enterprise search applications in a number of different organisations. Cleverley and Burnett at Robert Gordon University, Aberdeen, have published a number of papers on the way in which enterprise search has been used in a specific large oil and gas company. In particular in 2018 they identified that the root causes of search dissatisfaction were problems with the technology implementation, the quality of the content and the extent to which users were trained.

In 2019 the two authors published an excellent overview of enterprise search based on interviews with vendors and users.

Two other related areas where there has been a reasonable amount of research is the use of internal search applications by professional users, such as lawyers, patent agents, recruitment agents and clinical staff, and analyses of the way in which search is a core component in the successful completion of a task. In the case of professional users the research indicates that there are some significant differences in the use of specific features of the user interface.

Taking into account the research papers cited in the 2018 and 2021 studies it would indicate that there are probably fewer than 50 peer-reviewed papers into enterprise search-related topics despite the fact that millions of employees around the world use these systems to undertake business-critical searches. Each week there are around 200 research papers published in the IR section of arXiv but only a few have any relevance to the document-centric repositories that dominate enterprise information resources.

There are a number of reasons for the lack of interest by academic research teams in enterprise search. These include:

• The concern of organisations that the research will reveal its strengths and weaknesses
• During the duration of a typical three-year PhD study there could be very significant changes in business direction that might invalidate the research, for example an acquisition, divestment, or the establishment of a new line of business
• Because enterprise search is security trimmed to ensure that only employees with appropriate access permissions see certain information it is very difficult to know whether the inability of an employee to find information is actually an outcome of a security barrier
• The dominant ‘enterprise search’ application is Microsoft Search but this is an atypical example as it is specifically designed to work within the Microsoft technical architecture. As a result, inter alia snippets are not well-presented and the search analytics applications are very limited.
• There are currently no undergraduate or graduate courses in enterprise search technology and management in North America or Europe. This means that there is very limited knowledge of enterprise search within even the Information School community and no incentive to undertake search to enhance the visibility of a department. As far as the IT management community is concerned it is of note that neither the British Computer Society nor the Association for Computer Machinery in the USA has published a book on enterprise search.
• The standard academic career progression of PhD, post-doctoral research, lecturer and upwards to potentially Professor does not accommodate time spent in a corporate enterprise search role which would not count towards academic advancement.

Another issue that faces enterprise search managers is that there are no enterprise search conferences at which good practice can be formulated and shared. At one time there was an Enterprise Search Summit in the USA but this became just one of many tracks at the annual KM World conference held in Washington D.C. each year. An Enterprise Search Europe event was launched in 2011 but was discontinued in 2014. The primary reason was a lack of sponsorship support from vendors who felt that this was not a good use of the time of their sales teams.

The Information Retrieval Specialist Group (IRSG) of the British Computer Society does include enterprise search in the scope of its annual one-day Search Solutions Conference and there is an Industry Day at the European Conference on Information Retrieval (also managed by IRSG) but as with the Search Solutions Conference there is no specific focus on enterprise search.

References