General

Machines capable of conversation

Machines capable of conversation

Machines are developed that can talk and participate in a conversation

According to a Technology Review article published this week, a startup called Cognitive Code has developed software that it claims will allow common devices to talk to humans. This study could help manufacturers of some devices, such as mobile phones or toys, use technology to add the ability to talk to their products.

According to Leslie Spring, the company's chief technology officer, with these algorithms, instead of writing an email on a PDA, we could instruct the device to send an email to Tom and tell him that we'll be there in 10 minutes.

The problem the company is trying to solve is known as natural language processing, and it has been thoroughly investigated for decades by various internationally renowned laboratories. Some computer programs are already capable of parsing basic information from inputs that do not exactly match commands. Some well-known examples are chatbots like Alice and Jabberwacky, programs that simulate a conversation through text input.

According to Spring, the Cognitive Code product, called SILVIA, is more advanced than the previous ones for several reasons. First of all, SILVIA remembers and understands the context of a conversation. For example, if you are talking about the movie Star Wars and you ask what the plot is, the system searches previous chunks of conversation to find an explanation for the question about the movie plot, rather than giving a general definition of plot or the plot of another movie or book mentioned before Star Wars. Another critical aspect of SILVIA that makes it different, Spring notes, is its ability to understand concepts expressed in various ways and express unique, meaningful responses.

The system works like this: During a conversation, words are converted into conceptual data, Spring explains. SILVIA takes these concepts and mixes them with other conceptual data stored in short-term memory (information from the current conversation) or in long-term memory (information stored during previous training sessions). SILVIA then transforms the resulting concepts back into human language. Sometimes software can cause programs to run on a computer or perform other tasks required to interact with the outside world. For example, you could save a file, query a search engine, or send an email.

Source: Technology Review


Video: Creating Chatbots Using TensorFlow. Chatbot Tutorial. Deep Learning Training. Edureka (September 2020).