Abstract:

Fifty years have passed since scientific study of natural human languages gained prominence, with two distinct fields: theoretical linguistics, and the computationally grounded Natural Language Processing (NLP), the latter of which is our primary interest. Over this time, advances with NLP software and algorithms have been made, and presently there is a wide selection of especially finely niched research available. Fewer are full systems which unite pieces of this finely niched research for a shared purpose. I would like to introduce Ehlic, a user interactive system of knowledge exchange. Ehlic incorporates both open source code from the NLP community, and new code based on concepts in theoretical linguistics. The user interacts with Ehlic completely by means of the English language; the user may provide statements of knowledge and pose questions, and in the case of Ehlic's uncertainty, the user may be posed a question in return. The current system operates in a subset of the English grammar: interactions are conducted in present tense, with nouns, verbs, and adjectives, which may be compounded into phrases, relative clauses, and sentences. Words and phrases that represent concepts are manifest in the knowledge model, a semantic network, as either instance or class types. An initial reference to a concept introduces a new node into the semantic network, with any relationships represented in the mentioning statement encoded as typed links to other nodes. Subsequent references resolve to the same node, and new knowledge is incorporated into its vicinity. Subsequent (future) incarnations of Ehlic will continue to incorporate still more newly coded theoretical linguistics algorithms and open source community NLP software, increasing the capabilities of Ehlic as both a research tool, and eventually a working interactive knowledge exchange system.