16-01-2013, 12:55 PM
TERMINOLOGY MINING IN SOCIAL MEDIA
NEW MEDIA
REQUIRE NEW TECHNIQUES
Most of the communication in social media is in textual
form. While social media authors adhere to most rules of
text production, the low level of editorial oversight, the perceived
informality of the media, and the comparatively high
degree of interactivity create a new communicative situation.
There are no previous genres for this new type of communication
| new text | to model itself after: new conventions
for expression are created apace and we can expect several
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
NEW TECHNIQUES REQUIRE
RETHINKING OLD CONCEPTS
As argued in the previous section, social media constitute
a semantically volatile domain, and if we intend to operate
textually in such an environment we need to employ a
methodology that can re-align its semantic model according
to observed language use. A theoretical perspective that _ts
particularly well with this requirement is the view professed
by structural linguistics that words can be characterized by
the contexts in which they occur, and that semantic similarities
between words can be quanti_ed on the basis of
distributional information. This idea, most concisely stated
in Harris [3], has been enormously inuential and has been
operationalized by a family of statistical algorithms known
as word space models [19, 13], which include well-known algorithms
like Latent Semantic Analysis (LSA [8]) and Hyperspace
Analogue to Language (HAL [10]).