In the last five years, the gap between the capabilities of the human brain and that of a computer or other electronic devices has been decreasing at a great speed. Today there are systems that not only read information, but are able to understand and remember it. In this line, IBM developed Watson: an artificial intelligence computer technology platform that uses Natural Language Processing and Machine Learning technology to reveal knowledge from a large set of unstructured data.
The current relevance of this type of machine capabilities lies in the fact that, in recent years, thousands of devices, systems and applications have been created for multiple uses that have generated and generate millions of data. Due to their quantity and diversity, this data is impossible for humans to process in reasonable times. The vast majority of them are and are generated "unstructured". It is believed to be invisible to humans and machines and there is no one to process it.
"What cognitive systems seek is to amplify the capabilities that humans have in decision-making, information consumption or whatever," explained Serrana Casella, leader of BigData & Analytics at IBM. Within the Watson system, IBM has developed and already has available to the public its Bluemix cloud, where you can find all its applications with cognitive capabilities. There are more than 100 services linked to cognitive computing, internet of things and others will continue to be developed. An example of services with cognitive capabilities is Personality Insights: an application that tries to determine the personality of a person through a text input (a form of unstructured data), and places it in a ranking.
Watson also offers, in the Bluemix cloud, the Conversations service, through which chatbots or virtual agents can be created. These softwares combine machine learning, natural language understanding, and dialogue tools so that, in an integrated way, they generate a relationship with the user.
The uses of a chatbot can be very diverse. For example, any company can implement one on its website, which automatically responds to the most frequent queries from customers or users. This can also be applied to social networks and applications such as Facebook Messenger. Conversations can link the chatbot designed by any user in Bluemix with their Facebook page, for example.
Educate the robot
Inés Martínez, CEO of Second Act Creative Thinking explained to journalists the step by step to create a chatbot. The first thing to do is to establish certain intents or intentions. These are the ways in which a user could refer to a particular request or intention: How do I do? How can I?, among many other examples. Then you have to teach him the categories of knowledge (entities) that the user could make queries about. In the case of a chatbot for a bank, these could be: credit cards, accounts, savings bank, transactions.
With these parameters set, we proceed to create the dialogue, i.e. the structure of the conversation. The chatbot is tested directly and, each time it makes a mistake, the intents and entities are modified so that the chatbot learns and improves.
A use case of IBM chatbots is the one that the company Kona is already developing for Toc Toc Viajes. This travelbot will help improve the search for tourism packages for the agency's followers. Diego Cibils, co-founder and CEO of Kona, announced that this service will be available from January.
Source: Crónicas
Connect