Searching with Memex

Memex Structured Searching

Many organizations have structured databases. However, structured databases are not easy to search. This is because most search technologies, although powerful when conducting “free-text” searches, find structured information difficult to handle. This is particularly clear when structured databases are asked to search beyond a simple fielded structure to a search across relationships.
Unlike most of these “free-text” search engines, Memex search technology not only offers and excels in powerful ‘Search Engine’ style “free-text” searching, but does so while retaining both the structure of, and linkages between, an organization’s data. With Memex users can perform highly structured searches with the performance of a “free-text” search engine.

Memex Number and Date Searching

Searching for numbers and dates is fraught with challenges. Many search engines treat them as text i.e. words composed of digits and not even requiring specialized fields, despite their apparent differences.

With Memex search technology, numbers and dates are treated uniquely as numbers and dates, with all the range of searching that requires, even when they are embedded in text. Memex allows searching for ranges of dates and numbers, again within textual descriptions.

Memex search technology even considers the way data is recorded, so a search for “$127,000” can be found by a search query requesting data to be returned in the range “dollars and more then one hundred thousand but less than 200 thousand.”

Memex Intelligent Searching

In the world of intelligence professionals, there are frequent discussions as to whether exact searching or “fuzzy” searching delivers superior results. Indeed, searching for intelligence implies a constant struggle between the opposing parameters of precision and recall, as datasets grow, “fuzzy” searching returns increasing volumes of irrelevant data.

Memex intelligent searching resolves this dilemma, providing what is known as “hyper-complex Boolean” searching i.e. Boolean searching but without limitation to either the number of search terms allowed or the complexity of logic. Memex queries can easily run into thousands of terms.

Such queries are generated through “domain-specific” query expanders, capable of expanding a search to the best possible Boolean search. For example, a search for ‘Detroit’ as a place can also search for every  zip code, street and area in Detroit, whereas simply searching for “Detroit” as a name would search for misspellings, similar sounding variants and thesaurus driven look-ups for names with the same root (John, Jean, Ian, Jan, Jack etc.). This process also works algorithmically, so a search for a telephone number will look for hundreds of variants in terms of the way in which the telephone number has been written.

These “domain-specific” query expanders are created both in partnership with, and by Memex customers, ensuring that best practice in intelligent searching is captured and made available to every single Memex user. Users simply specify precise requirements at time of search. In short, Memex is not just an intelligent search technology, but an expert one, making use of the most intelligent searching available in the world today.