Will Expert System Deflate the Google Balloon with Semantic Technology?

Semantic search technology is the latest development in the web search world. It is fast gaining ground over its established rival, keyword search, which is the mainstay of Google’s search engine.

Expert System has launched Cogito, a semantic search program, which they claim will revolutionize web search. The company gained prominence in late 80s by supplying the language software to Microsoft for their spell-check software. Even though there are many players in the field of semantic search, Expert System is the frontrunner with Cogito.

Keywords are outdated

When users Google search with keywords, they get tens of thousands of results. Most of these are not relevant to the meaning of the keyword in that particular context. For example, when the word ‘rock’ is searched in Google, the results would bring forth links for the word meaning ‘stone’ as well as ‘rock music’ among others. The user has to sift through the irrelevant results to find the right one. This is both time-consuming and waste of energy.

Semantic search can avoid this by allowing secondary words connected to the keyword in the search. With this, the contextual meaning of the word becomes evident. Like ‘climber’ or ‘mountain’ for the meaning ‘stone’ and ‘band’ or ‘music’ for ‘rock music’. This gives pertinent and less number of results as compared to keyword search. The user can save time and effort with semantic search.

Google uses PageRank, a ranking algorithmic program for web pages to sort them in order of relevancy. Semantic search uses a lexical database with word meanings and their interrelatedness to zero in on its contextual meaning and to deliver precision results.

As keyword search is fast losing its position on the high pedestal, a number of players are emerging to occupy its place, other than semantic technology. Other search techniques coming up are linguistics and text mining, heuristics and ontology and statistical analysis. However, they also fail to comprehend the contextual meaning of the searched word, which leads to irrelevant results. Utmost these technologies can do is to understand the grammar and structure of the searched text.

Semantic Search vs Others

Semantic search is programmed to work like a human brain in understanding the text as it is meant to be. None of the other contenders has this ability. When a word with a variety of meanings is searched, the number of results would be mind-boggling for non-semantic searches.

Semantic search works by semantic analysis and by understanding the logic of the sentence. Semantic analysis helps in comprehending the contextual meaning of the words. This is known as ‘word sense disambiguation’. A word is ambiguous when it has different meanings in different contexts. For example, the word ‘fine’. To clarify the meaning of the word in a particular context, a semantic analysis of words around it is essential.

Other forms of searches are ill equipped to deal with the issue. They either ignore the ambiguous words and produce results or include all contextual meanings of the words and turn out millions of results.

Non-semantic search engines score nil when it comes to deep analysis. Each reacts differently when confronted with keywords that require deep analysis. A heuristic search of a sentence with two adjectives would show no results for those words as the search engine is not capable of processing them. Hence, they are neutralized and eliminated from the search.

Semantic Network – Its advantages

Expert System based on semantic network is confident of capturing the web search world, as other players are lagging behind in the technological race. The semantic network has the leverage of lexical database that helps in unraveling the labyrinth of word meanings and their intrasentential relationship. Webster’s Dictionary with its 350,000 words and 3 million interrelationships is stored in the database to make this possible.

The best feature of Expert System is its focus on common words, which is normally ignored by regular keyword searches. They concentrate only on specialized words and its meaning and neglect the common words in the search. In fact, common words constitute a whopping 90% of the searched content.

Semantic Web or Web 3.0

Expert System is leading the pack of players already targeting the Semantic Web pie. The contenders in the race include Microsoft-backed Powerset, Trovix and Yedda. Recently, Microsoft heated up the semantic scene, when they bought the data quality provider, Zoomix.

Until now, semantic technology was only a hyped-up theory. There was no validation to its claims being better than the presently used keyword search or other alternative technologies.

The main drawback of semantic technology is that it is not everyone’s cup of tea. It is complex and hard to build. As the technology is in its nascent stages, it is improbable to assume that foolproof software will be built in the near future. Semantic searches pose tricky performance issues and demand more processing cycles than its forerunners. Until these issues are sorted out, it is early days to predict its future.