It is quite common that new users confuse chatbot and search engine. Indeed, these two tools are quite similar and yet very different from each other.
In this article, we will discover what differentiates a chatbot from a search engine.
What interactions do users have with a chatbot and a search engine ?
As we explained in our article Chatbot vocabulary: terms to know, the term “chatbot” is a combination of the English terms “chat” and “robot” to designate “conversational robots”, also called “conversational agents”. These assistants are usually available as instant messengers on websites and in applications. Instead of talking to a human, users chat with a robot. As indicated in their name, conversational agents are based on the classic patterns of human conversations, one of the two interlocutors being the user, the other the robot.
This is the first real distinction between a chatbot and a search engine: chatbots reproduce as closely as possible the patterns of natural language conversations (human-human interactions) while search engines only allow users to obtain an answer to their query. In a search engine, there is no reproduction of a discussion between the machine and the user: a query brings up results. If the results are not suitable for the user, the user has to start a new query.
What are the differences between a chatbot’s databases and those of a search engine ?
The search scope of a search engine is much larger than a chatbot. Indeed, web browser search engines can probe thousands of websites, unlike a chatbot whose data has been previously selected and distributed in databases (or knowledge bases in Hubi.ai). However, as we know, the more data and the more domains are covered in the knowledge, the more frequent errors and confusions can be.
What role does the relevance of the user query play in the return of results ?
A chatbot displays only one result to the user, the one with the highest relevance score. Depending on how the chatbot works, several results can be displayed and it will be up to the user to select the answer that best fits his request.
Contrary to a chatbot, a search engine will return dozens or even hundreds or thousands of results. In these results, some will perfectly match the user’s query while others will be less suitable although correlated with the query. As a general rule, the results that contain all or most of the keywords in the query are displayed on the first page of results. At the end of the first page of results, or even on the second page, the results no longer necessarily contain all the keywords, but the subject matter may be close or simply correlated with the query. Although the results are sorted, all the results related to the user’s query are displayed.
Why are keywords so important in user queries ?
When a query is very specific, this implies two things: the longer and more specific the query, the more keywords there will be and the more difficult it will be to find a 100% match. The chances of finding the right answer are likely to be reduced. However, asking a query that is too short or too vague will not get an answer either since the number of results is likely to be too large.
So whether in a search engine or in a chatbot, it is important to ask queries that are neither too many keywords nor too few.
In concrete terms, how does the chatbot react to a real user request ?
Let’s take an example of a query related to “vacations”.
Whether our interlocutor is a machine or a human, a query consisting only of the word “vacations” will not allow him to understand what we are asking him, even if he will be able to deduce that our query is about vacations.
In a conversation between two people, the context given by the previous exchanges can lead the interlocutor to understand the purpose of our request. However, without a context, our interlocutor will need more details to be able to answer us correctly. He or she will probably ask us questions to determine the real subject of our request: “What do you want to know about vacations?”, “Do you want to take some time off?”, “What are your plans for your next vacation?”, etc.
With a chatbot, it’s pretty much the same thing. Depending on the way the chatbot works and the structure of its knowledge, there is a good chance that the chatbot will not find an answer either because there is no match in its knowledge or because there are too many results. If there are too many results, we can estimate that the chatbot needs as much precision as a human to find the right answer.
A search engine will display all results that match the query “vacations”. It will be up to the user to select the result that suits him best.
In conclusion, the difference between these two tools lies in the way the results are reported
Chatbot and search engine are therefore two distinct tools that respond differently to a similar need: the search for information. Whatever the medium used for a search, one must keep in mind that the structure and scope of the data from one tool to another is often very different. In addition, the way in which the query is formulated to interrogate this data is essential: it should be neither too precise nor too vague. Finally, the biggest difference between these two tools lies in the way the results are returned. Indeed, a chatbot will give only one answer while a search engine will display several. Rather than asking a succession of simple queries, a chatbot will reproduce a human conversation.