Apr 06

“TextWise Semantic Signatures® are based on vector spaces with thousands of dimensions, each corresponding to a single concept in some domain of interest. We use a special semantic dictionary to map the content of a given text document into a point within that conceptual space; and one can then gauge the similarity of two documents from the distance between their points in the conceptual space.”

The above quote is from SemanticHacker, the blog for a semantic search engine company called Textwise. Semantic search is the next generation of search, focusing on creating search engines that can take natural language queries and understand the relationships between the words in the query. It’s something that is difficult to grasp and equally difficult to explain as the quote above demonstrates* (and as my own attempt to explain also demonstrates). Yet it is an advancing technology that is going to have a big impact on our lives. Unfortunately, until someone figures out how to explain it in simple terms, it won’t have value.

As a professional communicator this is an issue I deal with everyday. In the semantic example, the best way to explain it would be to show an example, i.e.:

Type a request into the search engine: How do I make eggplant less bitter?

The engine understands the terms ‘how’ and ‘make’ mean there is a process involved. It understands that ‘eggplant’ is an object and ‘bitter’ is a characteristic of that object. Finally, it understands that the word ‘less’ is associated with bitter in a quantitative manner- we want less bitterness. After assembling all of this understanding it returns a search that has specific directions for making eggplant less bitter.

This is semantic search. Normal search just returns keyword combinations and phrases without this deeper understanding of context and the query: eggplant+bitter+less+how, etc. As a result you get more non-relevant results. Semantic search is an improvement in that it returns higher quality results.

In any business there is the need to explain the value of what you offer in very basic terms. I usually tell people that if you want tell your story in the most compelling manner you need to become your customer and then ask yourself:

What’s in it for me?

If you provide a strong answer to that question you will get your customer’s attention. If you don’t, you won’t.

*In fairness to Textwise, their posts are targeted to people with a very sophisticated understanding of semantic search. This just happened to be one of the more convoluted examples of a product description that I could find.

One Response to “Semantic Search: If you can’t explain it in simple terms it has no immediate value”

  1. Ron Kass Says:

    Techies explain tech in a technical way :) I guess thats why they don’t have social life? ;)
    the simple way to put their explanation in words is this..

    If we could place “things” (like web pages, or simple answers to questions) on a map (like the world map), where the location would be basically their “meaning” (what the post calls “points in the conceptual space”).. then two things with the same meaning would be closer to eachother.. like two neighboring countries.
    The closer two “things” are, the more relevant they are (conceptually closer).

    So the idea here is to place “things” on a “map” where “distances” can be measures.

    The tricky thing here is defining a “map” with its rules and measurements that actually have any coherent meaning to base the claim that “closer” means something real.

Leave a Reply