een Number Artst examined items and all of the remaining items. The system ExTracted Hearnes then makes recommendation of new items by ordering based on its high similarities with the selected items [5], [6 However, the content-based approach has the limitation such that it focuses on only the accessed items and is not prompt to immediate changes in the potential interest of Figure 1. The composition of individuals users. To overcome these limitations. we combine the content-based filtering approach and the genetic algorithm in A. Feature Extraction Phase ur proposed system. In this phase, we perform feature extraction to each B. Music feature extraction music track using CLAM; all of music track are formatted as The feature extraction is a technique that derives MP3(Mpeg-1 Audio Layer 3). The CLAM outputs properties from specific data, such as music, document and extraction results as XML files containing music features of photo. It is also each track. The system then parses the XMl files to generate for grouping static resources. In our initial individuals for IGA. Our proposed system considers proposed system, we employ the content-based filtering to five extracted features which consist of real numbers. Fig. 1 acquire information from music data. The analysis of items is shows the example of an IGA's individual composed of the sential step of filtering items in the content-based ng approach. We then use a feature extraction tool (i.e. extracted features M) to analyze the properties of items B. Evaluation phase As a software framework for research and application development on the audio and music domain, the CLaM The proposed recommender system grants its users the role of evaluation the fitness value of each music track. each provides complex audio signal analysis, transformations and user assigns his or her own rating scores to music tracks according to their subjective preferences. By this means of C. Interactive Genetic Algorithm the scoring metric, the users can represent their favor rating enetic Algorithms(GAs)are stochastic search methods to different recommended items. The recommender system inspired from the mechanism of natural evolution and evolves a population based on user evaluation data genetic inheritance. GAs work on a population of candidate C. Interactive Ga Phase solutions; each solution has a fitness value indicating its Interactive Ga phase is the fundamental component of closeness to the optimal solution of the problem. The our system since it proposes promising items (i.e. music solutions having higher fitness values than others are track) to the users based on their own evaluations selected and also survive to the next generation. GAs then Similar to the genetic algorithm, IGA also works on the produce better offspring (i.e. new solutions)by the basis of genetic inheritance and it has evolutionary operators combination of selected solutions. The methods can discover, (i. e, selection, crossover and mutation) preserve, and propagate promising sub-solutions [7],[ 8] In this system, we do not consider mutation because we Interactive Genetic Algorithm (IGA) is also an focus on finding items which are most appropriate to us optimization method as the genetic algorithm. In IGA however the fitness values of candidate solutions are based preferences. Since the mutation operator would cause candidate solutions to deviate from the common pattem on the evaluations of users according to their own discovered by the evolution process, it should omit. Fig. 2 preferences [9]. Our proposed system uses IGa to recognize shows how to operate Interactive Ga phase in this system user favorite since the user can judge the fitness value of each solution (i.e, music track). The user preference, thus, evaluation (i.e, Evaluation Phase) and the algorithm executes three separate steps (Selection, Crossover and IIL. SYSTEM OVERVIEW Matching)as shown in Fig. 2 1) Selection: We apply the truncation selection method The recommender system described in this pa n the genetic algorithms. The content-based this system, since the item having lower rating scores technique is applied to generate the initial population of hould not be considered to make new recommendation genetic algorithms. In the proposed system, we employ the elitism to impose high selection pressure to favor the top T% interactive genetic algorithm so that the users can directly evaluate fitness values of candidate solution themselves. Due (the constant variable T) candidate solutions having higher to the subjective evaluations, our system can recognize and itness values. The remaining items having lower rating recommend items tailored with different user preferences scores are then discarded [10], [11]. After the selection, half recommender system is divided into three ph of the selected items would be applied the crossover operator follow: feature extraction, evaluation and Interactive Ga in a probabilistic manner. phase http://www.clam-project.org Volume 6] 2010 2nd International Conference on Computer Engineering and Technology V6-415 5    =    <. 8          3    &    . < <   . @FA@1A3 .-   < 55                  5 5        5     3  -      .  <     < 55    55 3        8    9  - 55  5          53 (    5   3 (  55   .  5    <    9    3       5          <  553% 8 *33 $#+ =55 3   .  .&     55 -5             $# 5- 58        3> "  !   #   *+   5       -      3.& 55    G       -         5     5< 3    -     -        --   8   3   5 < 5  *33 .  + <   <   3   - 5- 555  < @2A@HA3 ( -    *(+    5 =       3 ( ( .-  -  <   -        . 5 @BA3755 ( =  -      I    -    *33 &+3   5   < 3 (((3 J !7D!D(!%    < 55<      3    <    9  55       55     3 (  55   .  5  -          - -   -3  <I--     =     . 5 3       -  5  .E     $     "   ! %#  > 5EKK...3 5I3 :>3  5  -     &#  (  5 . 5  8   & $#G &  LC *5>  # C+3  $# 5 8 M#     &3   5M#     -  (3 7 55    -8.   <3:3> .8 5 (; - 5 83   &#   55          -  - &3!       .      &   <I- 5 3         5  -         3      --55 < - 3 "  !&#  ( -  5        5         55 5     *33  &+< . - 3     (.&  <    - 5 *33 -   +3 (  .    <.        .   555   5 3     5 .       -     5 - <  - 5    3 :3 ? ..5( -5  3 (  55   ( .& <   - *33   &# +     8  5 5 *   $       #+.  :3?3 ('    )%55              -  .     <  & .   3 :        5 5-5*N *    -< *+     -     -3        -  .    @>/A@>>A3   .<55-5  5<<  3 [Volume 6] 2010 2nd International Conference on Computer Engineering and Technology V6-415