Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Possibility of applying fuzzy logic in the e-Learning system

2,381 views

Published on

Introduction
Algorithms
Soundex
Difference
Levenshtein distance
Model of applying fuzzy logic in searching
Word index
Fuzzy logic algorithm in searching
Further development

  • Be the first to comment

  • Be the first to like this

Possibility of applying fuzzy logic in the e-Learning system

  1. 1. Preaković, D., Remenar, V., Grgurević, I. Faculty of Traffic and Transport Sciences, University of Zagreb Vukelićeva 4, 10000 Zagreb, Croatia {dragan.perakovic, vladimir.remenar, ivan.grgurevic}@fpz.hr CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  2. 2. Keynotes <ul><li>Introduction </li></ul><ul><li>Algorithms </li></ul><ul><ul><li>Soundex </li></ul></ul><ul><ul><li>Difference </li></ul></ul><ul><ul><li>Levenshtein distance </li></ul></ul><ul><li>Model of applying fuzzy logic in searching </li></ul><ul><ul><li>Word index </li></ul></ul><ul><ul><li>Fuzzy logic algorithm in searching </li></ul></ul><ul><li>Further development </li></ul><ul><li>Questions </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  3. 3. Introduction <ul><li>Searching… </li></ul><ul><li>Closed system searching </li></ul><ul><li>Link-based relevance calculation </li></ul><ul><li>Finding relevant information </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  4. 4. Algorithms <ul><li>Large number of algorithms </li></ul><ul><li>Most used: Soundex, D-M Soundex, Metaphone, Levenshtein </li></ul><ul><li>All phonetic algorithms – English language </li></ul><ul><li>Algorithms used in this research: </li></ul><ul><ul><li>Soundex </li></ul></ul><ul><ul><li>Difference </li></ul></ul><ul><ul><li>Levenshtein distance </li></ul></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  5. 5. Soundex <ul><li>Designed and patented in 1918 </li></ul><ul><li>“ Sounds like” algorithm </li></ul><ul><li>Word encoding: </li></ul><ul><ul><li>first letter followed by three numbers </li></ul></ul><ul><ul><li>w, y and vowels are not encoded </li></ul></ul><ul><ul><li>numerical value assigned to characters </li></ul></ul><ul><li>Disadvantage: error in first letter gives completely wrong result </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  6. 6. Difference <ul><li>Same principle as Soundex algorithm </li></ul><ul><li>Ability to set weight value of “similarity” </li></ul><ul><li>Range from 1 to 4 </li></ul><ul><ul><li>1 wider results </li></ul></ul><ul><ul><li>4 most similar </li></ul></ul><ul><li>In this research value is set to 3 </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  7. 7. Levenshtein distance <ul><li>Calculation of the differences between two values </li></ul><ul><li>Inserting, deleting or switching the character places </li></ul><ul><li>Creating n+1 and m+1 matrix </li></ul><ul><ul><li>n – length of the first word </li></ul></ul><ul><ul><li>m – length of the second word </li></ul></ul><ul><li>Final “distance” is on the last place </li></ul><ul><li>in matrix d[n,m] </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 . p r o m e t 0 1 2 3 4 5 6 p 1 0 1 2 3 4 6 o 2 1 1 2 3 4 5 r 3 2 2 2 3 4 5 m 4 3 3 3 2 3 4 e 5 4 4 4 3 2 3 t 6 5 5 5 4 3 2
  8. 8. Model of applying fuzzy logic in searching <ul><li>None of the phonetic algorithms can be used for Croatian language </li></ul><ul><li>Finding what is needed requires a search engine with fuzzy logic </li></ul><ul><li>Word index is required – corpus with word “weight” </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  9. 9. Word index <ul><li>Some words are used more often </li></ul><ul><li>The more frequently used words have </li></ul><ul><ul><li>greater significance (“weight”, ponder) </li></ul></ul><ul><li>Every record is input </li></ul><ul><li>Special marks are removed </li></ul><ul><li>Common words are not indexed (“ili”, “ako”) </li></ul><ul><li>Word index contains 40,000 words </li></ul><ul><li>“ Top” words: “promet”, “prometu”, “sustava” </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  10. 10. Fuzzy logic algorithm in searching <ul><li>Search term is divided into separate </li></ul><ul><li>words </li></ul><ul><li>Each word is checked for errors </li></ul><ul><li>If word is mistyped, algorithm is </li></ul><ul><li>searching the correct word </li></ul><ul><li>Database is queried </li></ul><ul><li>Relevance is calculated </li></ul><ul><li>Results are ordered by relevance </li></ul>CEC IIS 2008 , Faculty of Organization and Informatics, Varaždin, 200 8 .
  11. 11. Further development and conclusion <ul><li>Algorithm is most efficient in specific areas as word index is required </li></ul><ul><li>Developed algorithm has proven to be more than satisfactory </li></ul><ul><li>Algorithm has the possibility to find words with the same root, e.g.: “sustav”, “sustava”, “sustavu” </li></ul><ul><li>Still, it is necessary to upgrade so it can recognize words as same like: “baza”, “baze” </li></ul>CECIIS 2008, Faculty of Organization and Informatics, Varaždin, 2008.
  12. 12. Questions? CECIIS 2008, Faculty of Organization and Informatics, Varaždin, 2008.
  13. 13. Thank you for attention! CECIIS 2008, Faculty of Organization and Informatics, Varaždin, 2008.

×