CHATBOTS

Mehmet Utku Pişkin

113200067

Mustafa Eymen Alpaydın

Raşit Deniz Cansever

The history of the Chatbots starts from 1960s as the MIT professor Joseph Weizenbaum
developed the first Chatbot ever. He called it ELIZA. It was using pattern matching and
substitution methodology for simulating the conversation. The idea was implementing a program
that can mimic human conversation. It was using a script that was simulating a psychotherapist.
Undoubtedly the script had enormous effect on natural language processing and unnatural
intelligence.

In the years that followed, Chatbot makers used Weizenbaum’s model to be more realistic in that
matter of being human-like interactions. In the year 1972 an American psychiatrist Kenneth Colby
developed a Chatbot named PARRY. Its purpose was simulating the disease. It was also a natural
language program like ELIZA that resembles the way of thinking of an individual. These two
Chatbots were the most influencing programs of the history of Chatbots. Later on other popular
Chatbots are developed in the second half of the 20th century.

In 2009 a software company called WeChat created a much more advanced Chatbot in China.
When it launched it got popular really fast and it conquered the hearts of the users of the
program. Today it’s a really popular thriving social media platform.

WeChat was actually really important in the history of Chatbots because it made an easy way to
create Chatbots. That’s why it became to be one of the most favored ways for marketers and
employers to enormously reduce the work because they are interacting with the customers
online.
The first wave of artificial data technology came in early 2016 in the design of Chatbots. One of
the most used social media platform Facebook was created and opportunity for developers to
build a chatbot for their trademark or a service. The reason behind that was that for the
customers could carry out some of their daily actions from inside their messaging platform.

Today Chatbots are widely used in following areas:

 E-commerce bots
 Content delivery
 Agenda/Scheduling
 Conference bots
 Handling minor daily tasks
 Event reservation
 Personalized helpers
 Travel bots
Mathematical Foundations of Chatbots and Their Approaches

There are four crucial steps to creating a Chatbot. These are;

 Maxim of Quantity: Saying only what is not implied.

Acknowledging that Chatbots are replying with prefixed sentences, recurrent neural network
encodings were used as vectors to make bots answer accordingly to the given sentences by
identifying them as agents in the created network. These are considered as purposeless
mimicry agents as well. On the other hand we have Intention-Based agents. Google Home,
Apple’s Siri or Microsoft’s Cortana can be given as examples to these. Intention-Based Agents
have two main problems to solve in understanding a given sentence.

 Identify what the user wants the machine to do. This is called “the intent”.
 Understanding the intent so that the machine can take required actions.

This process is held and managed by NLP as bot takes the given sentence and it parses the
sentence to words. Thus, allowing the bot to take actions by reverse building the sentence after
analyzing and understanding the parsed words. Lastly, we have Conversational agents. The
difference between these agents and Intent-Based agents is that Conversational agents use the
same way with turning these sentences to multi-turn conversations by keeping the track of the
state of the conversation. They must also understand when the person wants to switch the
subject. We can give Jarvis, the assistant of Tony Stark from the movie series Iron Man.
Repetitive dialogues with agents can also effect cognitive biases as they converse.
Examples for Real World Applications

Chatbots became digital friends. There are so much success example for real-word applications.
First of all Microsoft’s Little Ice, and Xiaolce in Chinese, became a social media phenomenon.

Xiaolce is Microsoft’s biggest Chatbot and also one of the most successful story of the Chatbots
for Microsoft. Also Xiaolce would be the most technically sophisticated bot in the world.

Secondly Mitsuku is the one of the popular bot. Even Mitsuku was founded in 2005, and is a
relatively old chatterbot, that keeps winning prizes for her engaging conversation. Mitsuku does
not serve any specific purpose, all its aim is engaging and entertaining users. She has so much
awards based on their ability to have human-like conversation.

Also there is an interesting Chatbot as Replika.ai. Replika asks a potentially endless barrage of
questions and follows people’s social media (if you allow it) to understand people enough to be
an engaging conversationalist.

These are the Chatbot to conversation actually. There are digital assistants as Chatbot too.

Firstly, Alexa. Even Alexa is the voice bot; it became a best bot. Skills in Alexa terminology is
applications which allow Alexa to complete certain voice tasks. It is difficult to compare
Chatbot’s skills however it is safe to say that with its vast developer community Alexa is more
skilled bot than any other Chatbot. She can help people shop, listen to music, run pools or get
people’s kids to play karaoke.

There is a good assistant chat bot as FacebookM. It is on of the most successful bot that people
use are tightly integrated into their activities. So tightly integrated that people don’t even
notice that.

The other popular assistant chat bot is Google Assistant. Available in all Android phones with
the swipe of the screen, Google Assistant is a holistic digital concierge. Additionally, assistant
can answer questions and learn about users to offer them personalized news or suggestions.