Information Retrieval with Human Indexing

Improving Information Retrieval
with Human Indexing (continued)

Why then are online help file indexes often of poor quality? In most cases, it's because the help author has computer generated the index. Which brings us to another point. How should the index be created?
In reality, computers cannot generate indexes, they can only generate concordances. That is, they can create a list of all the words found within a document. Again, we are faced with the same problem we had when discussing search engines.

Mind is as mind does

Once more, we can look to how book indexes are created and see what we can learn from the publishing industry. Major publishing companies that produce quality reference and trade books use human indexers. These individuals are professionals at analyzing content and developing indexes which are designed to serve the precise needs of information seekers.
Professional indexers read through each and every line of content and ponder over which possible words would best convey the meaning of each paragraph within the text. They add essential terms to the index, even if the text does not use these exact words.
In this way the index user can find the precise information that is being searched for. Additionally, subcategories are added to the main topics, allowing the user to decide which aspect of the main topic is most relevant to his or her search.
But won't it be more expensive to contract a human indexer? Not necessarily. Think about the minutes that add up to hours during which employees fruitlessly try to locate needed information on the company intranet. What if they need the information to answer a question in real-time, for instance for a support call? What if they cannot find the information at all?
Intranet search engines, while low in initial cost, have a hidden and very significant continuing cost as they drain time and productivity from a company.

Index Interfaces
There are many user interfaces for such indexes. Some prefer a back-of-the-book style index, with alphabetic headings at the top of the page and between letter groups. Upon clicking on the letter, the user is brought to the appropriate letter group.
Others prefer an online help style index like that used in software, mentioned previously. In such an index, as soon as the user begins typing the first letters of the word, the index automatically scrolls to the appropriate letter and to the exact word as he continues tying. This is referred to as type-ahead scrolling. Which is better depends on the needs of the company, and the design of its intranet.
Documents stored on a company intranet can have enormous value, but only if they can be found. Companies with intranets and intranet developers should consider the full range of options when determining how they supply access to such information.