Document comparison with a weighted topic hierarchy

Abstract

Abstract

Natutal language processing: Perspective of CIC-IPN

“…The frequencies of the words can be propagated up the hierarchy: say, one text 23 times mentioned animals (cows, lambs, and sheep), and the other text 22 times (cats, dogs, and parrots); even if specific animals are different, the texts are likely to be on similar topics [81]. Note that in our method the topics are not just categorized (animals, books, etc.)…”

Reacting to Unexpected Events and Communicating in Spite of Mixed Ontologies

“…It uses a concept tree. For this reason, it can find an article talking about shoes, even if the article does not contain such word, but it contains instead boot, moccasin, sandals..., even if these words may refer to other contexts or concepts: moccasin is also a tribe of American Indians... [3,4] extend [9] by using weights for selecting concepts.…”

“…3,4 The LIA world contains resources, global variables (time, temperature...), regional variables, 5 agents, scripts, and unexpected events. Agents and scripts are constructed using LIA [15].…”

“…Finally, some of the threads stopped in (3) are never restarted, due to lack of resources. Replanning 4 (not discussed here) may be needed. Execution of (2) to (8) Reaction thread, written in LIA.…”

“…To know which of the normal threads stopped in (3) are continued in (5), in spite of the unexpected event, the agent uses an incompatibility algorithm (not described here) to detect which threads can not run simultaneously with the reaction thread (4).…”