Survival of the Fittest – Utilization of Natural selection Mechanisms for Improving PLE

Download Survival of the Fittest – Utilization of Natural selection Mechanisms for Improving PLE

Post on 10-May-2015

958 views

Category:

Education

3 download

Embed Size (px)

TRANSCRIPT

<ul><li>1.Survival of the Fittest Utilization of Natural selection Mechanisms for Improving PLE <ul><li>Behnam Taraghi,Christian Stickel, Martin Ebner </li></ul></li></ul> <p>2. http://ple.tugraz.at Mashup of widgets 3. Darwins theory </p> <ul><li>- Survival of the fittest </li></ul> <ul><li>- Selection, Variation </li></ul> <ul><li>- Macro evolution </li></ul> <ul><li>- Micro evolution </li></ul> <p>4. Selections </p> <ul><li>Evolution theory of natural selections: </li></ul> <ul><li><ul><li>- Reproduction rate </li></ul></li></ul> <ul><li><ul><li>- Mortality </li></ul></li></ul> <ul><li><ul><li>- Population size </li></ul></li></ul> <ul><li><ul><li>- Environmental capacity </li></ul></li></ul> <ul><li><ul><li>- Cycle of updates, replacements &amp; new widgets </li></ul></li></ul> <ul><li><ul><li>- # widgets </li></ul></li></ul> <ul><li><ul><li>Max # widgets on UI &amp; # users </li></ul></li></ul> <p>Different probabilities for the survival are the base for theselection mechanism . 5. Selection Mechanisms* Stabilizing Selection : - Favorites the average. - Decrease of variability within the population.</p> <ul><li>Disruptive Selection : </li></ul> <ul><li>Directed against the average. </li></ul> <ul><li>Splits the population into new species. </li></ul> <ul><li>Directed Selection : </li></ul> <ul><li>Works against individuals on one side of distribution. </li></ul> <p>*Solbrig O.T. (1970). Principles and Methods of Plant Biosystematics. The Mac-Millan Company. Collier-Mac Millan Limited, London. *Solbrig O.T. &amp; D.J Solbrig. (1979). Populationbiology and evolution. Addision-Wesley. Publ. Co. Reading Mass. 6. r/K Selection Theory* Tradeoff btw.quantity&amp;qualityof offspring In long term K-strategy is superior. Quality succeeds in long run over quantity. </p> <ul><li>r-Strategy </li></ul> <ul><li>Succeeds in unpredictable, unknown environments. </li></ul> <ul><li>High reproduction rate short lifespan </li></ul> <ul><li>K-Strategy </li></ul> <ul><li>Succeeds in predictable, known environments. </li></ul> <ul><li>Constant growth, ruled by population density &amp; env. capacity </li></ul> <ul><li>Usually close to the max. capacity </li></ul> <ul><li>Slower adaptation longer lifespan </li></ul> <p>InPLEa mixed approach was applied. *Pianka E.R. (1970). On r and K selection. American naturalist 104, 592-597. *MacArthur, R. and Wilson, E.O. (1967). The Theory of Island Biography, Princeton University Press (2001 reprint). 7. Variations </p> <ul><li>Shift in genotypes or generic sequence </li></ul> <ul><li>Mutation : </li></ul> <ul><li><ul><li>Random process aiming at generation of new alternatives </li></ul></li></ul> <ul><li><ul><li>Ex. change in DNA structure </li></ul></li></ul> <ul><li><ul><li>Continuously happening </li></ul></li></ul> <ul><li>Recombination : </li></ul> <ul><li><ul><li>Not random process </li></ul></li></ul> <ul><li><ul><li>Combining &amp; distributing genetic materials (DNA, RNA) </li></ul></li></ul> <p>The Evolution never stops. InPLE : slight update of existing functionality or UI InPLE : combining code of different widgets to build new ones 8. Tracking module in PLE </p> <ul><li>Tracks users behavior on widgets </li></ul> <ul><li>Deep retrieve of statistics data </li></ul> <ul><li><ul><li>Frequency of widgets usage </li></ul></li></ul> <ul><li><ul><li>Features used in each widget </li></ul></li></ul> <ul><li>Realized via IWC </li></ul> <ul><li>- App. 1000 users registered up to now. </li></ul> <ul><li>- App. 30% active users </li></ul> <ul><li>- Top 5 most used widgets out of 30: </li></ul> <ul><li>- tugWidget, tccourses, tugllBlogs, mail, changeThemeColor </li></ul> <ul><li>Top 5 most activated widgets: </li></ul> <ul><li><ul><li>weatherForcast, RSSReader, twitter, TUGLibrary, leoDictionary </li></ul></li></ul> <ul><li>weatherForcast &amp; newsgroup are improved according to K-strategy </li></ul> <p>Most activated widgets are not necessarily most used ones 9. Conclusion &amp; Future Work </p> <ul><li>Tracking module helps </li></ul> <ul><li><ul><li>To get knowledge about user behavior </li></ul></li></ul> <ul><li><ul><li>To get user preferences </li></ul></li></ul> <ul><li><ul><li>To categorize different user groups </li></ul></li></ul> <ul><li><ul><li>Improve the PLE with variations &amp; selections </li></ul></li></ul> <ul><li>Missing qualitative data </li></ul> <ul><li><ul><li>Qualitative rating system is needed in PLE </li></ul></li></ul> <ul><li><ul><li>Small feedback questionnaire for each widget</li></ul></li></ul> <p>10. SOCIAL LEARNING Computer and Information Services http://tugraz.at http:// elearning .tugraz.at Slides available at:http:// www.slideshare.net/behi_at b.taraghi(at)tugraz.at </p>