See: Doszkocs, T. (2010). Semantic Search Engines Mean Well. Online (Weston, Conn.), 34(4), 36-42.
In 2010, when Dosczkos wrote this article, semantic searching was already being incorporated into Bing and Google searches. In his words, “Semantic searching transfers the burden of thinking up synonyms and related terms from a human to a computer.” For example, if a person searches for “library organization system” the semantic search engine would understand that to mean classification system, and would show results related to that term.
Dosczkos gives a brief history of early search engines such as AltaVista and InfoSeek. These search engines had some “behind-the-scenes” methods to help return more relevant results, from stop words to inverted indexes. The problem with these was that they lacked the human element – the semantic analysis – which would distinguish between two very different search queries with the same words, such as police state and state police.
When Google introduced its Page Rank algorithm in 1997, it greatly improved the relevance of search results, and internet searching became easier, quicker, and more accurate. I can remember the dawn of this type of searching when first using AskJeeves.com, now simply Ask.com. The idea was that you could type a full question into the search box, and get a good answer, instead of doing a keyword search and muddling through a page of links. I remember feeling strange asking the internet a question, but at least I was getting answers!
Dosczkos explains that search engines are now trying to grapple with our natural language, which is both meaningful and ambiguous. The challenge is to deal with polysemy, synonymy, user intentions, context, disambiguation, and more. It hasn’t always been easy for search engines to accomplish, but Google and Bing are both making strides toward improving the experience of searching.
He further explains some of the methods Google and Bing programmers surely use to improve semantic searching:
- Utilizing existing semantic resources, from thesauri to dictionaries to classification systems
- Natural language processing, which seeks to understand language through a variety processing tools
- Metadata or faceted clustering
Not only are Google and Bing integrating semantic search technologies into their search engines, there are specialized search engines that are doing the same. PubMed, Web of Science, and Lexis Nexis, among other, incorporate semantic search technology. The benefits of this are clear: searchers are getting better results, quicker, and without all the hassle. But is it really beneficial? To answer this question, I will examine Nicholas Carr’s article, “Is Google making us stupid?” for a future review.
Finally, one statement that Dosczkos made sticks in my mind. He states that semantic searching is what librarians are after as professionals – finding meaning in a user’s query. Reference librarians are the mediators between people and information: at the reference desk, we’re tasked with interpreting a patron’s question into a meaningful search that will retrieve relevant results.
If search engines become experts at semantic searching, will librarians no longer be needed? I don’t think so – you can never really replace a human being with a machine when it comes to navigating the vast world of information, or conducting complex research. Further, the machines are only as good as their makers.