Nice introductory article of Steve Rubel on the invasion of advertising in the blogosphere. The idea is simple and old as the world: look at what the consumers do (i.e., how they categorise a certain product in this case) and use the system to advertise the products of the company you are working for. But what’s in a category, a tag? How the category people choose relate to their background, to their needs?

The ConceptNet knowledgebase is a semantic network. 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.

M. Minsky. Commonsense-based interfaces. Communications of the ACM, 43(8):67–73, August 2000. [url]

———————————-

The core of this paper is to describe the principle of the commonsense-based approach to AI: that in order to achieve some intelligence from the system, we need to equip the sistem with adequate knowledge in order to enable the machine to represent the same problem in several ways.

This principle is the key of the project developed a the MIT Media Lab called “The OpenMind project”, which idea is to constitute a database of the common bits of notion that everybody knows and that follows under the definition of CommonSense.

Marvin Minsky has made many contributions to AI, cognitive psychology, mathematics, computational linguistics, robotics, and optics. In recent years he has worked chiefly on imparting to machines the human capacity for commonsense reasoning. His conception of human intellectual structure and function is presented in The Society of Mind (CDROM, book) which is also the title of the course he teaches at MIT.

[url]

The author describe some technique to implement an ontology of tags:

The goal of this ontology is to model the relationship between an agent, an arbitrary resource, and one or more tags. This relationship is embodied in one or more taggings, which are temporal events associating the actors.

CellTrack is free and open source utility for tracking the location of GSM cellular phones. Currently C500 smartphones running Windows Mobile are supported.

A free CellTrack client can be downloaded and installed on your phone. This runs in the background and collects information on the GSM cell towers in your phone’s vicinity. This information can be used to estimate the geographical location of your phone.

A bunch of position papers for this nice workshop! Worth a read.

SCIgen is a program that generates random Computer Science research papers, including graphs, figures, and citations. It uses a hand-written context-free grammar to form all elements of the papers. Our aim here is to maximize amusement, rather than coherence.

Three approaches most interest me in the current research I am doing:

1. Cognitive Semantics in Formal Ontologies

This particular research puts the accents on the definition of the activities related to the particular formalization used to support it: “Supporting human beings in geographical space requires ontologies that are developed paying attention to both objects and activities [Kuhn, 2001]. In addition, concepts to be added to a certain domain ontology have to be grounded in order to represent the common usage.

2. Pragmatic Relevance

Relevance theory states that an input is relevant when, and only when, its process-ing yields a positive cognitive effect [Wilson et al., 1995], meaning that the receiver of the input will generate and act an expectation of a particular cognitive effect to be achieved by the  incoming input [Matsui, 2001]. These ‘acts’ of expectation may be observed and accounted as yielding back to the reasoning and inference process of the learner and on the particu-lar perspective s/he is taking in relation to the processed input.

3. Computational cognitive modeling using Agents

We can use computational cognitive modeling to support the categorization process of the messages posted in space. This will require the definition of the proper ontological definition/position for each message. Additionally, the agent can highlight messages connected to the user’s interaction on the system. Here we can reuse existing metodology of Frank [2000].

S. Spaccapietra, C. Parent, N. Cullot, and C. Vangenot. On using conceptual modeling for ontologies. In Proceedings of the International Workshop on the Intelligent Networked and Mobile Systems, Ontologies for Networked Systems (ONS) track, located at the 5th International Conference in Web Information System Engineering WISE04, Brisbane, Australia, 22nd-24th of November 2004. [url]

————————————

This paper offer a new approach for using Conceptual Modeling in ontologies instead of simple Description Logic. The idea is to use the Extended Entity-Relationship (EER) models, which are seen as offering the richest semantic expressiveness. This technique is seen as an advantage for humans and for Agents as well, because they can explore a conceptual schema without going trough a long list of DL axioms.

An interesting research direction for the data structures is said to be the spatio-temporal extension of DL languages.