Imagining Affiliation Rules for Content Mining.


74 views
Uploaded on:
Category: Funny / Jokes
Description
Imagining Affiliation Rules for Content Mining - Sangjik Lee Pak Chung Wong, Paul Whitney, Jim Thomas Pacific Northwest National Lab Presentation An affiliation standard in information mining is a ramifications of the structure X - > Y where X is an arrangement of precursor things and Y is the resulting thing.
Transcripts
Slide 1

Imagining Association Rules for Text Mining - Sangjik Lee Pak Chung Wong, Paul Whitney, Jim Thomas Pacific Northwest National Laboratory

Slide 2

Introduction An affiliation standard in information mining is a ramifications of the structure X - > Y where X is an arrangement of precursor things and Y is the resulting thing. For quite a long time specialists have created numerous instruments to imagine affiliation rules. On the other hand, few of these instruments can deal with more than many standards, and none of them can successfully oversee rules with various antece-gouges. In this way, it is to a great degree hard to picture and comprehend the affiliation data of an extensive information set notwithstanding when every one of the tenets are accessible.

Slide 3

Association Powerful information examination system that shows up every now and again in information mining writing. An illustration affiliation guideline of a general store database is 80% of the general population who purchase diapers and infant power likewise purchase child oil.

Slide 4

The framework was produced to bolster content mining and perception research on huge unstructured report corpora. The center is to contemplate the connections and suggestions among themes, or expressive ideas, that are utilized to describe a corpus. The objective is to find vital affiliation rules inside of a corpus such that the vicinity of an arrangement of points in an article infers the vicinity of another subject.

Slide 5

For instance, one may learn in feature news that at whatever point the words “Greenspan” and “inflation” happen, it is very presumably that the share trading system is likewise specified. Show the outcomes utilizing a news corpus with more than 3000 articles gathered from open sources.

Slide 6

Current Technology Two-Dimensional Matrix

Slide 7

Current Technology Directed Graph

Slide 8

Current Technology Directed Graph This strategy functions admirably when just a couple items(nodes) and associations(edges) are included. An affiliation diagram can rapidly transform into a tangled presentation with as few as twelve principles.

Slide 9

A Novel Visualization Technique To picture numerous to-one affiliation rules Instead of utilizing the tiles of a 2D grid to demonstrate the thing to-thing affiliation tenets, utilized the network to portray the guideline to-thing relationship.

Slide 10

A representation of thing relationship with backing >= 0.4% and certainty >= half

Slide 11

A Novel Visualization Technique ( Continued ) the grid\'s lines floor speak to the things (or subjects in the connection of content mining) the segments speak to the thing affiliations. The blue and red pieces of every section (standard) speak to the precursor and the guideline\'s resulting. The things\' personalities are appeared along the right half of the network. The certainty and bolster levels of the principles are given by the comparing bar graphs in diverse scales at the most distant end of the framework.

Slide 12

A Novel Visualization Technique - Advantage There is for all intents and purposes no maximum point of confinement on the quantity of things in a forerunner. We can examine the affiliation\'s dispersions rules (level hub) and in addition the things inside of (vertical pivot) all the while. the personality of individual things inside of a predecessor gathering is plainly appeared. Since all the metadata are plotted at the far end and the sections\' tallness are scaled so that the front segments don\'t obstruct the back ones, couple of impediments happen.

Slide 15

Conclusion and future work Applied the new method to a content mining framework to dissect a substantial content corpus. The outcomes show that our outline can without much of a stretch handle many different forerunner affiliation rules in a 3D show. Long haul objective is to coordinate a large number of apparatuses and methods into a solitary representation environment that gives time arrangement investigation, theory clarification and report rundown.

Slide 16

References Pak Chung Wong, Paul Whitney, and Jim Thomas. Envisioning Association Rules for Text Mining . In Graham Wills and Daniel Keim, editors, Proceedings of IEEE Information Visualization \'99, Los Alamitos, CA, 1999. IEEE CS Press Pak Chung Wong, Wendy Cowley, Harlan Foote, Elizabeth Jurrus, and Jim Thomas. Picturing Sequential Patterns for Text Mining . Procedures IEEE Information Visualization 2000, Salt Lake City, Utah, Oct 8 - Oct 13, 2000. Nancy E. Mill operator, Pak Chung Wong , Mary Brewster, and Harlan Foote. Point ISLANDS - A Wavelet-Based Text Visualization System . In David Ebert, Hans Hagan, and Holly Rushmeier, editors, Proceedings IEEE Visualization \'98, pages 189 - 196, New York, NY, Oct 18 - 23, 1998. ACM Press.

Slide 17

.

Recommended
View more...