A merger of (at least) four disciplines. A merger of (at least) four disciplines

On top of this, the same text may be rephrased according to different conventions:

• On top of this, the same text may be rephrased according to different conventions:

• 5pm, 17:00, 5 o'clock in the afternoon …
• April 1, 2006, 1/4/2006, 4/1/2006, April Fool's Day, …
• Adelaide, ADL …

• As a result, techniques must allow for ambiguity and imprecise text.

• May use a resource such as WordNet - a semantic network resource which comprises separate networks for nouns, verbs, adjectives and adverbs with the basic element being a set of synonyms (synset).

• May also use "edit distance"

• This level facilitates the removal of common typographical errors caused by transposition (eg Flinders entered as Flidners), omission (eg. Fliders) or insertion (eg. Flindters).

Applications include:

• Applications include:

• Summarisation of documents
• Automatic classification of documents into categories
• Automatic detection of categories (clustering)
• Outlier document detection
• Pattern analysis for providence
• Change detection
• Clique identification

The extraction of the names of people and companies that occur in newspaper articles of a given topic, say wireless technology, to try to infer who the players are in that field.

• The extraction of the names of people and companies that occur in newspaper articles of a given topic, say wireless technology, to try to infer who the players are in that field.

• In genomics - which proteins interact with which other proteins. This has been approached by looking at which words co-occur in articles that discuss the proteins in order to predict such interactions.

• Automatic classification of newspaper articles according to the words used in the article.

We can also combine a number of algorithms by feeding the results of one into another.

• We can also combine a number of algorithms by feeding the results of one into another.

• This changes the semantics of the rules generated - effectively the rule generated discusses a set of other rules…

• The strength of the association between A and B is increasing.

Explorative Analysis

• Explorative Analysis

• no initial data hypothesis
• iterative search for structures

• Confirmative Analysis

• initial hypothesis
• goal-oriented examination of hypothesis
• confirmation or rejection of hypothesis

• Explanation

• facts to be presented are fixed a priori
• choice of appropriate visualisation technique
• high quality presentation of the data presenting the facts

Geometric

• Geometric

• Visualisation of geometric transformations and projections of the data
• Scatter Plot Matrices, Landscapes, Parallel Coordinates

• Iconic

• Visualisation of the data values as features of icons
• Chernoff Faces, Stick Figures, Shape Coding

• Pixel Oriented

• Hierarchical

• Visualisation of the data using a hierarchical partitioning into subspaces
• Dimensional Stacking, Worlds within Worlds, Treemap, InfoCube

• Graph Based

Each attribute represented by a single value Coloured pixel.

• Each attribute represented by a single value Coloured pixel.

• Each attribute is presented in a separate window

Filtering

• Filtering

•