Please enable JavaScript.
Coggle requires JavaScript to display documents.
Method (ENSEMBLE-BASED SIMILARITY (Training a Cluster Ensemble, Combining…
Method
-
EXPERIMENTS
Examine which sampling and modeling methods, and what classification algorithms perform well for the classification-based approach.
-
Gain insight about these approaches by analyzing the weights that the similarity metric learning approaches assign to each feature-specific similarity.
Problem Definition
-
the goal is to partition this set of documents into clusters such that each cluster corresponds to all documents that are associated with one event.
1 consider social media document representations using each individual feature, according to its type (e.g., textual or time data).
2 list the key types of features we extract from so- cial media documents, and define individual similarity met- rics for these feature types.
-
define a similarity metric for each feature, in a way that is appropriate for the feature’s domain
considered traditional text processing steps such as stop-word elimination and stemming, and examined the effect of each of these with respect to the individual textual features.
represent values as the number of minutes elapsed since the Unix epoch (i.e., since January 1st, 1970) and compute the similarity of two time/date values t1
and t2
-