|
"A program library based on the Natural Query Language
."
After eight years of research and development, AGIR introduces
NQL, a new technology that allows you to integrate a powerful text search engine
in your software solutions. Featuring the use of a natural language, NQL breaks
through the technical limitations of Boolean logic and solves a number of problems
related to the processing of large volumes of information. NQL incorporates
several new concepts that improve the performance of full text and structured searches.
NQL is a C++ programming library available for full text search. It uses a very
powerful algorithm that allows users to query the database using a natural (or common)
language, no boolean operators are required.
The engine uses a statistical analysis to sort the documents in order or relevance. To
sort, it uses factors such as presence of the terms, automatic proximity, concentration of
information, order of the terms, words weight...
NQL uses a new proprietary technique to create a sequential index which
give you these advantages:
The Naturel Query Language makes it easy for users
Our sorting methods (using 5 relevance factors) delivers a very precise answer
and saves a lot of time and system resources.
The indexing process isn't affected in a significant way by the size of the
database (in the long run, NQL will be much faster than competitive products).
The database Update is incremental (so, the updating process is very fast!).
NQL allows multiple users to access the indexes at the same time, without
affecting, the speed of the search.
The search is not only easy and precise, it is also very fast.
This new technology is very easy to integrate.
THE NATUREL QUERY LANGUAGE
How does Naturel handle multi-word searches
in comparison with traditional products?
Naturel is radically different from traditional text search products. Naturel does not
implicitly add "and," "or," or proximity operators between words. The
technology developed by our R&D team studies the "closest proximity" first,
then the "and" values and finally the "or" values. Naturel does the
work instead of asking the user to do it. Naturel also applies other statistical criteria
to increase the accuracy of results before displaying them. Naturel's cutting edge
technology lets users formulate queries in natural language.
After performing a statistical analysis of the results, Naturel displays the information
based on its relevance. AGIR's unique technology makes locating information easier
and corrects a number of imprecisions inherent to Boolean logic.
Results are sorted according to the following statistical criteria:
Presence
If all words in a query are present in a part of the text, that part is automatically
considered to be more relevant than sections where some of the words are missing.
Proximity
Naturel searches through a document, finds words in the proper sequence and then calculates
the words' proximity. The closer the words are to each other, the greater the importance
assigned to the document that contains them.
Concentration
When looking for a particular word, expression or sentence, Naturel assigns greater
importance to texts whose main subject relates to the information the user is looking for.
Order
A text that contains words in exactly the same order as in the query is considered more
relevant than texts that contain the words in a different order.
Weight
Naturel weights words according to the number of times they appear in the database. Words
that occur often are considered less important than those that appear less frequently.
For example:
If the user is looking for Article 409 in a legal text, Naturel presents the document that
has the most information on 409 before it presents documents that contain many instances
of the word article. Why? Because 409 appears less often in the database than article does,
Naturel gives more weight to 409. In fact, a word that occurs many times in a database loses
importance, as do stop words.
Here is an example of how Naturel differs from traditional technology
Query: The user is looking for a candidate who is fluent in English,
French and Chinese. The candidate should have experience in
international politics and an educational background in law, finance
and marketing. A knowledge of computers is also required.
Note: The perfect candidate may prove hard to find!
Traditional
technology:
| The
first attempt: |
"English
& French & Chinese & International & Law &
Finance & Marketing & Computers." |
Result: The query
is too restrictive and would not yield a result
The
second attempt:
|
"English
or French or Chinese or Politics or International or Law or
Finance or Marketing or Computers". |
Result: The query would most likely present all the candidates
in the database.
If you were looking for the best candidate, what would you do? Without
extensive knowledge of the text retrieval system, it would be very
difficult to narrow the search using queries like these.
Naturel
technology:
The query can be expressed in natural language.
| "Candidate
who speaks English, French, Chinese, has experience in
international politics, law, finance and marketing, and is
computer literate." |
Using this algorithm, Naturel will find that no document matches the
criteria perfectly, but that three documents may be of interest.
The first candidate satisfies most of the criteria, but does not speak
Chinese or have any marketing experience.
The second person speaks all three languages, but does not have
experience in law or in international politics and is not computer
literate.
The third candidate speaks English and has experience in law.
The documents are presented in the order of their relevance. A given
document is always more relevant than the document that follows it.
The Boolean operators normally used in traditional products are also
available in Naturel.
| Products |
Contact Us
| Download | FAQ
| Home Page
| Français
|
Copyright © 2000 AGIR, all rights
reserved. |
