Article Keyword Videos to Watch
Internet
Click on the image to start the video.
|
Related Topics
Images - Links - Articles
Washington
Related Images
|
Frequently Answered Questions
Every self respecting webmaster knows that if you offer a service on the web you have to give some sort of customer service. One of the most common solutions is a Frequently Asked Question page (FAQ). FAQ pages are easy to create and implement, and are very cost effective. As a result of the wide spread use of FAQ pages many developers have developed FAQ engines. An FAQ engine has an input form for the webmaster to input the FAQs and a search box to help the user find the question he needs answered. Most of these engines are quite good and help both the web master and the end user.
Many websites, even with the help of the FAQ engine, still have a hard time helping the users find the question they are looking for. This is caused largely due to the fact that regular search engines, that search based on keywords, are not efficient for Question/Answer searches. Question/Answer searches require a deeper understanding of both the question in the database and the question being asked in order to match the question with the appropriate answer.
To accomplish this task, which is referred to as "question mapping", an intelligent linguistic engine and a special subject related Knowledge Base are needed. The linguistic engine breaks the sentence down to its basic structure and then refers to the Knowledge Base to figure out what the user means. Although the intelligent linguistic engine can break the sentence down to its core meaning, the engine still requires these special knowledge bases, also known as ontologies, to fully understand the question.
The intelligent engine can understand this question: "where can I find a good pair of black shoes?" and break it down into its basic argument/predicate structure. While the basic understanding of the sentence is complete the engine still doesn't know what a "pair of black shoes" are, the engine understands that the user is looking for the pair of shoes but it doesn't have the knowledge of what the shoes are. Much like a non English speaker that has learned to speak English, the engine understands the sentence structure but doesn't know what everything is. This is where the ontology comes in, the ontology knows what black shoes are and can pass that information on to the intelligent engine. With the understanding and knowledge the engine now fully understands what the user has asked and can find the most appropriate answer. At this point we can change the name of our FAQ pages from Frequently Asked Questions to Frequently Answered Questions, being that now we can really answer the question being asked.
Many companies have tried to implement some or all of these intelligent technologies, some with more success and some with less success. But one thing all these companies have in common is the lack of a stronger syntactic engine to break the sentence down into its argument/predicate structure. Most of these companies try to fill the gap with technologies like pattern matching. Pattern matching, while quite efficient for small sentences (2-5 words), lacks the dynamics of understanding when it comes to larger sentences (6-13 words). Large sentences contain too much information for technologies, such as pattern matching, to handle. In order to understand a sentence containing more than five words a stronger syntactic engine is needed.
Although several companies have tried to build a strong syntactic engine over the past two decades, as of today none of them have successfully introduced one to the market. Linguistic Agents Ltd (est. 1999) of Jerusalem, Israel (www.linguisticagents.com) has successful developed a strong syntactic engine that breaks the sentence down almost the same way the human brain breaks sentences down. These technologies are years ahead of the tradition natural language technologies available on the market today.
With Linguistic Agents' Technologies, developing an intelligent question mapping application becomes simple. All of the core technologies needed come neatly packaged in one ultra light program (500k with all supporting layers). All that is left to do is compile a knowledge base, i.e. Ontology, type up the questions and answers, and add a search box to the existing FAQ web page. The user can then type in a question and Linguistic Agents' engine will extract the meaning, match it with the appropriate answer and return the result to the user.
Life can be easy as a web developer, all you have to do is find the right technologies and then everything begins to move smoothly.
About the Author: Sruly Taber is the CIO of Linguistic Agents Ltd. Linguistic Agents has developed intelligent linguistic technologies.
|