Artificial intelligence, personal assistant and chat robot and other technologies are rapidly rise. With the development of Siri, Alexa, Cortana and Google (Weibo) assistant and other products, personal assistants that can connect to the Internet and Internet of Things are becoming more common.
In 2017, almost all chat tools and smartphone operating systems offer chat robots or individual assistants. Although such personal assistants can argue as a true artificial intelligence still there is still controversy, almost everyone will use such a virtual assistant in the upcoming world.
If the developer wants to be active in this regard, fortunately, a series of services have appeared on the market, which can help develop custom artificial intelligence, chat robots, and personal assistants to achieve a variety of different purposes, such as connecting smart home equipment. Control DIY's media center, publish all kinds of information through personal intelligence assistants.
Due to the API (Application Interface) and the existence of these services, developers have a lot of choices. In 2015 to 2016, these services have developed rapidly. In 2017, developers will get more tools than ever. Currently, developing their own solutions is no longer difficult.
The following is the current part of services that help developers connect their independent applications and Internet access to voice recognition, chat robots, and artificial intelligence services.
01, wit.ai
WIT.AI provides developers with speech recognition and machine learning. The service can convert a voice command to text and understand these voice commands by accepting training. Wit.ai also provides some form of machine learning mechanism, you can train wit.ai to understand the commands that have never been seen. However, this is not automated. At the beginning of 2015, Wit.ai joined Facebook and opened the entire platform to public and private uses. Since then, WIT.AI has been developed in disconnection. However, the team put forward a great plan in 2017.
Wit.ai contains two main elements, users need to set in applications, which are "intent" and "instances". The definition of intent is an operation corresponding to the command (for example, turning on the light). The definition of the instance is a specific object or a message. Our artificial intelligence needs to know these objects and information to perform intentions (for example, which light bulb? Is it a smart light bulb? Does the bulb support specific color?) User does not need from head Started to create intentions, Wit.ai's developer community has provided a lot of intent to choose from.
Wit.ai also proposes the concept of "role", thereby distinguishing different instances in different environments (for example, numbers in different parts of the command can represent different meanings, such as age, order, and counting, etc.). Wit.ai has built some instance types, thus facilitating understanding, such as temperature, link, email, and duration.
Wit.ai has recently provided new features called "Story", and users can define typical conversations in a new way. You can set the initial problem, such as "How is Sydney Today?" Subsequent steps, as well as issues should be made. There is also a concept of "branch" here. When the system does not get all the necessary information, the conversation will be advanced in different ways.
Wit.ai provides an API for iOS, Android, Node.js, Raspberry, Ruby, Python, C, Rust, and Windows Phone, which provides a JavaScript plugin for front-end developers.
02, API.ai
Api.ai is a chat robot API, and the function is similar to wit.ai, and also the concept of intent and instance. Its machine learning ability can guess whether the command statement used by the user is slightly different from the strategy of hardcoding to the artificial assistant. In September 2016, Api.ai was acquired by Google, which has become one of the main ways of developing a session interface on Google Home platform.
One of the main differences between API.Ai and Wit.ai concerned is "domain". The domain is the entire knowledge and data structure from API.Ai, which can be used for each API.Ai agent (in API.ai, the application is called "agent"). The domain may include knowledge consisting of ordinary verbs and content types. For example, it can understand data, such as the "Booked Restaurant" and "Booking Hotel". This also integrates a lot of information on the encyclopedia, including historical, word definitions, important characters (such as stars, writers and movie roles), movies and share prices.
Api.ai can be used free, but the information on its website is confusing. By 2016, API.ai is not completely free. Most "domains" require you to upgrade your account during use, but the specific price is unknown (developer needs to contact API.ai sales team). Api.ai also provides a paid enterprise version, allowing the entire service to run on the private cloud, and get more service support. This is valuable if strict requirements are used in private environments.
Api.ai offers Android, iOS, Apple Watch, Node.js, Cordova, Unity, C ++, Xamarin, Python, and JavaScript SDK (Software Development Kit). By integrating Unity, API.AI can also be used on other platforms not listed above. In addition, this can also be integrated with Amazon Echo, Skype, Slack, Facebook Messenger, and Microsoft Cortana.
03, Melissa
If you want to make human intelligence programming, it is a fans of the Raspberry Pie, so you can look at Melissa. Melissa is an open source personal assistant, written in Python, can run on platforms such as Raspberry, Windows, OS X, and Linux. Melissa's update is very frequent.
Melissa provides a voice control function that is always open, and a series of dialogues are integrated by default, including record, constellation, definition from Wikipedia, and play music. If Python developers want to get complete control, then Melissa can do. Melissa's main developer Tanay Pant (Tanay Pant) introduced MELISSA's details in a book, which is also a detailed documentation for Melissa.
04, Clarifai
Clarifai is a completely different service that identifies pictures and video content. Clarifai has an autonomous depth learning engine that continues to optimize as users. If you want to develop your artificial intelligent prototype products to a new height, a good choice is to make it recognizes the object. Clarifai can complete photo tags, search similar photos, and inappropriate photos such as tagging. You can even use your own example data, and guide the platform to learn the new concept.
If you want to integrate it into your application, the Clarifai provides a REST API that can be integrated into common language, and on the other hand also provides Python, Java, and Node.js APIs. Clarifai's free service supports up to 5000 times a month.
05, TENSORFLOW
If you really want to develop artificial intelligence from the beginning, then Google's Tensorflow may be the best choice. Although Tensorflow is relatively troubles, you can better learn deep learning and artificial intelligence from it. Tensorflow is "an open source software library for numerical calculations using data flow map", which is best at training independent image identification system or language processing tool. With special data sets, you can also develop session manual intelligence with Tensorflow. For example, the Speakeasy API is a chat robot using millions of comments on the Reddit website.
There is almost no restrictions on projects that can be carried out by Tensorflow, and even developers have written new scripts to well-known American dramas "Old Friends".
in conclusion
There are a variety of services and APIs on the market, which can provide artificial intelligence, personal assistant and chat robots. Integrating these features to your own application does not require your level to achieve the height of computer scientists.
And if you are very confident to yourself, you can try Tensorflow directly to develop a refreshing new feature.
This article originally address: https://www.eeboard.com/news/siri-alexa-cortana/
Search for "Love Bo.com" to pay attention, daily update development board, intelligent hardware, open source hardware, activities, etc., you can make you master. Recommended attention!
[WeChat scanning picture can be directly paid]
Technology early know:
Looking back, see the design of these mobile phones is really bold.
Exposure: Nut Pro Super Configuration Double Pieces Remove Headphones
A new generation of Windows systems heavy exposure: global tagging
Predicting memory and solid state hard drives have finally cut prices
Intel has been lonely: save the motherboard to see AMD
Our other product:
