I am a newbie with chatbot and AI but is it possible if i share my idea here. The idea is to make a chatbot with search engine transplant in it. When user input the topic, the chatbot search it on internet and take the hotest one and then just take the core of the topic. What i mean is to make a chatbot that can update with every news on the net so the user can chat with easy and can gain information too altogether.

If it is possible, what algorithm should i learn?

thanks for your time.
I am sorry for the grammar mistake.

Hello, Rayga, and welcome to chatbots.org.

Your idea is certainly possible, but I think you’re going to find it to be very challenging for a number of reasons. Here are two of the biggest challenges, in my opinion:

The first challenge will be to make sure that your chatbot sounds like a chatbot (or, preferably, a real person) instead of a search engine. Chatbots that use search engines generally tend to read like the first entry of a Google search, and that’s not really a good way to describe a chatbot, so the “hottest” search result should be re-phrased in order to come across as conversation, rather than as a search engine result. There’s not yet (to my knowledge, at least) any one specific algorithm available to do this, and will probably require some rather creative coding.

Another challenge is the actual integration of your chatbot and a search engine. There are several search engine API’s available, depending on the language you choose for your chatbot, with some API’s being easier to use than others. Finding the right one for your needs will be a matter of research and experimentation.

There are/will be other things, too, that will prove difficult, but without knowing more information, it’s very hard for me to guess at what they might be. I hope that this helps you, in some small way.

Ah, i see.
Thank you Dave. Finally i see any reason why i should give it up for now.

Sorry for the late introduction. I am a college student. This idea is for my final project. Even though lot of people say exactly what you said. It’s all about how you change the “hottest” to be conversation. But i believe there is a way until i found this site. And you give some light to me.

I think i will find another one but i will need your help because it’ll be about chatbot again. It’s interesting in some way.

Seriously, i thank you.

But maybe if it is possible. I want to know. If i take this one for my final project. Is it possible to do it in 1 year?

Hi RICH does this, but its broken into (2) different “lobes”.

If you ask RICH “Who are the Beatles” and there is no local memory, it searches a list of Knowledge bases in an attempt to find an answer that is close to being a synopsis. If if cant find one, it attempts to create a synopsis from more detailed KB entries. (See Daves entry above on “humanizing” the response)

However if you ask RICH “Do you LIKE the Beatles, it uses the Google API to search through several preset news and other reference sites, and builds a list of web references. Then it “reads” (parses) the HTML removing the tags to build a text set. Then it parses these text sets looking for keyphrases and keywords that are already stored and categorized as to emotion and intensity.  By “reading” reference material on a subject , RICH can decide what it “feels” about that subject.

Doesnt always work the way you expect. The other day it came back and said that it “Liked” war, and “Yes. Probably.” liked Nazis. My wife went ruunning down the hall screaming “I knew it, its the Terminator” LOL (Truly hard to communicate just how funny this was)

To implement this type of capability you need a method that parses HTML and removes the tags. In Pseudo code it looks like

find “<” as position1
find next “>” as position2
make substring1 as 0 to position1
make substring2 as position2 to end of string (HTML)
concatenate substring1 and substring 2 as new HTML
call recurslvely until end of string

If what your trying to do is have a bot that “reads” the news so that it converse on events of the day, you might want to consider using an RSS instead of the search API. RICH reads the news using http://rss.cnn.com/rss/edition.rss. (Im having problems with the webapp version loading the RSS headlines into a dataset directly that I havent tracked down yet, but it works in the desktop version) Here you have the headlines and links to articles in xml format, and they are updated regularly.

You should probably tell the group what base Bot engine (if any) your using to see if there are specific modules that you can use for your particular circumstances, or if you are custom coding what language(s) are you using

VLG

wow… i’m glad to be here.
Thank you Vincent, i already took a note of what you said. And LOL there is some funny story too you have there.
I will try RICH that you recommended as a base later on. Seems like it’s too early to give up yet.

There are lot of great people here. I really am glad to be here and learn about chatbot like i’m in the middle of a class.

So, what should i learn next if i want to make this chatbot on smartphone?

Well, learning how to program for the Android and IOS operating systems may be a good starting point. I’m not sure if Apple has an SDK, but Android sure does.

Yeah, i choose Android because there are lot of people using it in my country and i’m in the middle of learning it.
what should i learn for the method or algorithm to make this chatbot ?

I am sorry beforehand if i look like a total newbie, because i really am.