Frequently Asked Questions about

the Open Mind Initiative

What is the Open Mind Initiative?

The Open Mind Initiative is a collaborative framework for developing "intelligent" software using the internet. Based on the traditional open source method, it supports domain experts (who provide algorithms), tool developers (who provides software infrastructure and tools) and non-specialist "netizens" (who contribute raw data).

What are the goals of Open Mind?

After many decades of research, there are still very many tasks for which computers are far worse than humans: recognizing speech, reading printed or handwritten text, recognizing objects from their image, understanding scenes, making complex plans, summarizing a story, and so on. It is clear that if such software could ever approach human performance, it would be extremely useful. Even if such software merely enhanced or aided human cognition, it would be useful. Much early research in "artificial intelligence" concentrated on small "toy" problems, such as game playing. There is a growing realization that we now need information contained in very large data sets.

• The principal goal of Open Mind is to develop "intelligent" software, in part by collecting such large data sets and providing an open infrastructure where different ideas can be tried out. The data and resulting software are made available to all.
• Another goal is to educate the public as to the problems in cognitive science, computer science, pattern recognition and related fields.

How does Open Mind differ from traditional open source and the Free Software Foundation?

The most important difference between traditional open source and Open Mind is that Open Mind relies on collecting, and exploiting large sets of data, such as the identities of millions of handwritten characters and spoken words, the names of objects in photographs, common sense about the world, and much, much more. This information is provided by non-expert "netizens."

Traditional open source	Open Mind
minimal use of netizens	netizens crucial
expert knowledge (e.g., C++filt, gdbm, Linux device drivers, ...)	informal knowledge (e.g., knowledge of speech sounds, written character identities, ...)
machine learning irrelevant	machine learning essential
web infrastructure useful	web infrastructure essential
most work is directly on the final software	most work is not directly on the final, but instead on infrastructure, data collection, data, etc.
hacker culture (roughly 10,000 contributors to Linux)	netizen and business culture (roughly 100,000,000 people on the web)
software released	software and data released

How does Open Mind different from data mining?

Data mining is the task of finding structure in existing (generally static) databases.

Data mining	Open Mind
type of data may not be available for the project desired (e.g., optical character recognition data is not available on the web)	the type of data is tailored to the project desired (e.g., characters are generated and presented over the web)
goal is usually interpreting structure in data that already exists	goal is usually building a classifier or other intelligent system
no interactive queries, thus slower learning and occassionally a lack of ambiguity resolution	interactive queries, thus faster learning and resolution of ambiguities
relatively fixed amount of data	new data encouraged

How can I participate?

If you'd like to be on our general mailing list, click here and put "subscribe openmind-general <your e-mail address>" in the body of the message. If you'd like to provide programming or other assistance to an ongoing project, click here. If you'd like to propose a new project (see also the list of suggested projects).