Search-chatbot that does hunt regulating Google

96 views Leave a comment

Sometimes we need not usually a series of links yet a sold answer to a request. The Google association is really modernized in this respect.

Ok Google and a built-in partner in a Google Allo follower concede to find answers to many questions, yet not to all. In this connection, Endurance starts building a possess bot on a basement of a Google API. Besides a Google API,  we use other solutions as good to weigh a exactness of a response.

What is a Google Web Search API?

The Google Web Search API lets we put Google Search in your web pages with JavaScript. You can hide a simple, energetic hunt box and arrangement hunt formula in your possess web pages or use a formula in innovative, programmatic ways.

REST, or Representational State Transfer, in a JSON/Atom Custom Search API is rather opposite from normal REST. Instead of providing entrance to resources, a API provides entrance to a service. As a result, a API provides a singular URI that acts as a use endpoint.

You can collect formula for a sold hunt by promulgation an HTTP GET request to a URI. You pass in a sum of a hunt ask as query parameters. The format for a JSON/Atom Custom Search API URI is:

Three query parameters are compulsory with any hunt request:

  • API key– Use the key query parameter to identify your application.
  • Custom hunt engine ID– Use either cx or cref to mention a tradition hunt engine we wish to use to perform this search.
    • Usecx for a hunt engine combined with the Control Panel
    • Usecref for a linked tradition hunt engine (does not ask for Google Site Search).
    • If both are specified,cx is used.
  • Search query– Use the q query parameter to mention your hunt expression.

For some-more sum we might visit:

The problem to be solved

The existent problem: Google does not always give accurate answers. It can give a series of links with illusive answers instead of a scold answer.

Our idea is: to make a available and fit chatbot  that will give genuine answers, not links.

As we see, Google infrequently does not give a applicable answer. Image credit: Endurance

Why is it important?

To hunt for something a is compulsory to use Google hunt or  a web browser. Our resolution works in any communicator: Messenger, Twitter, Slack, Gmail, Telegram + web interface.

It creates a hunt easier and provides a improved entrance to a information we need. A user gets not usually a series of links yet a sold answer to a sold question.  Our chatbot looks for answers regulating Google API, afterwards chooses and gives a some-more applicable answer. Besides, a chatbot parses a couple given by Google and answers on a basement of what it has found in a given link.

What is a advantage of a chatbot?

  1. Our chatbot looks for answers yet not for links with illusive answers.
  2. The chatbot is means to use analogies for a hunt for answers to questions
  3. It can review and investigate texts by approach of parsing than is, it can give a full answer to a question.
  4. It works in any follower with a open API.
  5. It learns, examining a hunt formula (if a user gets a applicable answer to a question).

What is an additional ability of a chatbot?

One can mention and supplement sum to a query.

For example:

You write on Google: «What is a continue like  currently in Moscow?»;

Google gives an answer.

If we write after that “And in San Francisco?”, Google will be incompetent to answer, since your new doubt lacks a word “what is a continue like”. Google does not remember a context.

The Endurance chatbot does remember a context and understands that we are referring to a weather, so it can answer your doubt “and in San Francisco?”   Holding a context is a categorical advantage of a chatbot’s intelligent search.

Plans to customize a chatbot

We devise to exercise a following options: “continue a search”, “specify”, “give some-more details”, “point out what is some-more important”, “learn”, “continue as a content goes”.

Smart Search

The simple judgment of a intelligent hunt is that a hunt bot can also use not usually a approach request, yet also “analogies and synonymic row,” i.e, a chatbot analyzes and searches for meaning.

The hunt is finished not usually by keywords, yet also by synonyms.

For example, a hunt content lacks a keywords compulsory for a answer. To capacitate a chatbot to give an answer in a deficiency of a preferred keyword, we devise to supplement a process of analogies; that is, a hunt bot will be means to find a compulsory answer, regulating analogies, even if a content lacks a compulsory keyword.

What is to be finished on a project

The Endurance plan skeleton to urge a peculiarity of answers by approach of charity to a users a some-more available channel of a information hunt and gaining.

You can exam a bot on Telegram:

On a website:

Send your feedback to

Written by George Fomitchev, Founder of Endurance

Comment this news or article