Text Analysis Techniques
30th May 2006
|
Ever wonder how the search engines decide which website to list in the top positions of the SERP’s? Of course you did…and do. That’s part of the daily thoughts of an SEO. 1) Summarization: This technology can summarize an entire document into a sentence, or develop keywords based on the concept of the document. 2) Categorization: This is a type of text analytics that will decipher what type of document it is (doc, html, pdf, white paper, research, etc), the subject matter, the source of information (mass media, educational government, corporation, etc), type of industry, geographic location, and date of publication. 3) Entity Extraction: This form of Text Analysis will discover people names, company names, location names, job titles, product info, company revenue, etc. 4) Latent Semantic Analysis: This is a technique used for information retreival. Latent Semantic Analysis identifies the overall subject matter of documents by mainly identifying synonyms and differentiating polysems (same word with multiple meanings). Currently, today’s search results don’t improve a whole lot when you type a long query….like a full sentence that is a natural sounding question. However, I would expect one of the Search Engines to master all these text analysis techniques very soon. This will change the way SEO’s need to think about keyword planning, and website optimization. For example, it might be very possible in the near future to type a natural sounding question into the search box, and get exactly the correct answer. At that point, SEO’s will need to think about all the details and numbers on a webpage, not just keywords. Optimizing for Keywords, and link building will still be very important. However, detailed content will make a big difference. Related posts: |
















May 30th, 2006 at
In the end - it always comes down to content… relevant content!
April 3rd, 2007 at
I will have to agree with you stephen. Content is and always will be King.
Im going to dive into the topic of Latent Semantic Analysis though because it seems very interesting. Nice article btw Shimmon.