Where Is ConceptNet, Watson?

I was unfamiliar with ConceptNet 4, crafted by the Massachusetts Institute of Technology (MIT), until I saw a news story, "Top Artificial Intelligence System Is as Smart as a 4-Year-Old," on the Computerworld website. (See: computerworld.com/s/article/print/9240801/Top_Artificial_Intelligence_system_is_as_smart_as_a_4_year_old.)

The point of view of the article is that four-years-olds and the MIT "smart" system are about equal on the playground. Robert Sloan, a professor and head of computer science at the University of Illinois at Chicago, is quoted as saying: "We're still very far from programs with common sense and artificial intelligence that can answer comprehension questions with the skill of a child of eight."

Sloan evaluated the ConceptNet 4 system and found that it was on a par with a four-year-old with average intelligence. The downfall of the ConceptNet 4 system was in the area of "common sense."

My experience with smart software is that the systems are often trapped in their own rules and knowledgebase. Stepping outside of those constraints is difficult for most of the smart software to which I have been exposed. Google's smart software taps into a range of fancy mathematics and deep pool of knowledge in its log files. Algorithms react to changes in certain values, giving the system the appearance of intuitive behavior. Common sense? That is not a phrase I associate with most vendors of software that invoke the buzzword "artificial intelligence."

The ConceptNet research project has been in the works at MIT for more than 10 years. According to archived content about the project: "The ConceptNet knowledgebase is a semantic network presently available in two versions: concise (200,000 assertions) and full (1.6 million assertions). Commonsense knowledge in ConceptNet encompasses the spatial, physical, social, temporal and psychological aspects of everyday life. Whereas similar large-scale semantic knowledgebases like Cyc and WordNet are carefully handcrafted, ConceptNet is generated automatically from the 700,000 sentences of the Open Mind Common Sense Project—a World Wide Web-based collaboration with over 14,000 authors."

The description continues, "ConceptNet is a unique resource in that it captures a wide range of commonsense concepts and relations, such as those found in the Cyc knowledgebase, yet this knowledge is structured not as a complex and intricate logical framework, but rather as a simple, easy-to-use semantic network, like WordNet. While ConceptNet still supports many of the same applications as WordNet, such as query expansion and determining semantic similarity, its focus on concepts rather than words, its more diverse relational ontology and its emphasis on informal conceptual connectedness over formal linguistic rigor allow it to go beyond WordNet to make practical, context-oriented, commonsense inferences over real-world texts." (See: http://web.media.mit.edu/~hugo/conceptnet.)

Things computers should know

With each release of ConceptNet, the semantic system has been refined and fine-tuned. After years of effort, MIT has achieved the intelligence of a four-year-old. That seems harsh for a project nurtured in MIT's Media Lab.

Now at Version 5, according to the ConceptNet website, it is described as follow: "ConceptNet is a semantic network containing lots of things computers should know about the world, especially when understanding text written by people. It is built from nodes representing concepts, in the form of words or short phrases of natural language, and labeled relationships between them. These are the kinds of things computers need to know to search for information better, answer questions and understand people's goals. If you wanted to build your own Watson, this should be a good place to start!"

One major enhancement in the current version is beefier "hypergraph" capabilities. "Graph search" has become a catchphrase. The idea is that the relationship between and among concepts or entities are calculated, used and sometimes presented to a user. MIT says: "ConceptNet 5 is a graph. To be precise, it's a hypergraph, meaning it has edges about edges. Each statement in ConceptNet has justifications pointing to it, explaining where it comes from and how reliable the information seems to be. Previous versions of ConceptNet have been distributed as idiosyncratic database structures plus some software to interact with them. ConceptNet 5 is not a piece of software or a database; it is a graph. It's a set of nodes and edges, which we can represent in multiple formats."

The relationships forged by ConceptNet's system can be displayed to make connections among concepts easily discernible. The problem is the phrase "smart as a four-year-old." I think the phrase suggests that systems like ConceptNet and Watson are not yet ready for prime time.

Watson is IBM's marketing poster child for the $100-billion-per-year company's text analytics, search and big data capabilities. Watson information is available from the deep and frequently updated Watson subsite. You can work through a large amount of information. I clicked on a link inviting me to download the "Watson Op Ad."

The description of Watson on the IBM website is interesting as well: "Cognitive systems like Watson may transform how organizations think, act and operate in the future. Learning through interactions, they deliver evidence-based responses driving better outcomes. (See: http://goo.gl/mKy2f.)