Wouldn’t it be nice if everyone just understood what we said?
In a perfect world, our communication would be coherent and unambiguous, letting us provide our thoughts to others without a filter. That is absolutely not the world that we inhabit.
We must interpret the meaning of the writings and vocal communications of others, often on the fly through our own filter of comprehension and past experiences. On top of that, we have homonyms, homophones, sarcasm and metaphors that can take years to learn the contextual meaning of.
There is so much ambiguity in our language that the fact that we as humans are able to parse it, and derive meaning from this word soup, and reply with contextual understanding is absolutely mind-boggling. Just ask anyone who is working in the field of NLP (natural language processing) how much goes into making Alexa understand that you want the lights lowered in the living room for movie night.
The activity of speaking to our digital assistants and having them find us a fancy Thai food restaurant with decent reviews may not seem like a big deal, but it is. There is a reason why Alan Turing used the interpretation of natural language as a benchmark in his test to determine if a computer possesses artificial intelligence. The sheer complexity of the human language is staggering and the ability of modern computers to process it is one of the most daunting challenges faced by scientists.
The ability of computers to properly understand our queries and provide us actual meaningful answers to our questions hinges on the integration of semantics into our computers processes and searches. Semantics can be defined as the study of meaning. It can therefore be taken that a semantic search means to search with meaning.
To illustrate what a profound leap this has been, we can look to the old ways that search engines used to operate. Typically, the search engine would use a lexical system to determine what results to give you, meaning that it would look for literal matches to the query words typed by the user. The results given only contained those key query words, with no understanding of the actual question being asked.
Helpful, sure, but it’s a bit too close to a worldwide Dewey Decimal system. If you need help finding a book on learning how to talk to your teenage daughter, you go and ask the librarian.
Thankfully, through the development of semantic searches, we can now ask questions to our computers in the same way that we would ask a librarian and have them interpret the actual meaning of what we are asking. The queries that we make are being interpreted by countless algorithms that parse our words and organize them such that the computer can effectively “understand” the intention behind the words themselves. Our results are better and we more often get what we initially wanted.
Just as an aside, KISSPlatform uses a semantic search to help you easily and quickly find patents – contact us for more information. Quite frankly, we are a bit spoiled by this, in the same way that the automobile replaced the horse.
Now that’s not to say that things are perfect (as anyone using a search engine will tell you), but we are getting closer to that place where we can get a direct answer to our queries and no longer need to trade convenience for diligence. However, semantic searches can also be used to find other answers to our queries that were not initially considered as possibilities by the user, and this can be a powerful tool as well. The more you know how to properly use semantic search, the quality, relevance and expansiveness of your results drastically improve.
That’s why in our upcoming webinar, we will be looking at some of these semantic search engines (like ours) and how to use them to your advantage and search with meaning. Check it out on June 1st, 2022, I hope to see you there!
Potthast, M., Hagen, M., & Stein, B. (2020). The Dilemma of the Direct Answer. ACM SIGIR Forum, 54(1). https://webis.de/downloads/publications/papers/potthast_2020j.pdf
Bast, H., Buchhold, B., & Haussmann, E. (2016). Semantic Search on Text and Knowledge Bases. Foundations and Trends® in Information Retrieval, 10(1), 119–271. https://doi.org/10.1561/1500000032