COLLABORATIVE FILTERING

COLLABORATIVE FILTERING

Introduction

Predictor score is considered to be a type of tool that is used when calculating and predicting the possibility of business decisions plus financial conditions. There are several methods used for uniting predictor score, this paper discusses and highlights collaborative filtering. Collaborative filtering is a methodology that has two aspects which are narrow sense and general sense.

Collaborative Filtering

Collaborative filtering is an approach utilized in generating automatic predictions about various interests that expert requires making business decisions. It entails gathering of preferences information from various users in collaborations (Koren, Y., & Bell, R., 2015). The main assumption in this approach is that if individual A shares a similar opinion with individual B regarding work issue there is a possibility that individual A sharing individual B opinion concerning different issue than an opinion from a randomly chosen individual.

Advantages

1. Collaborative filtering structures function by individuals who are present in the system and it is anticipated that individuals to be good at evaluation of information rather than calculated function (Koren, Y., & Bell, R., 2015).
2. Collaborative filtering does not need extraction and content analysis
3. It functions excellently with complicated objects such as multimedia. This approach is independent of machine-oriented representations of items being recommended.
4. Collaborative filtering is more diverse and is considered to be a serendipitous recommendation

Disadvantages

1. Data sparsity- several business-oriented systems are structured on large data sets, therefore the user-object matrix which is utilized in CF, therefore, is sparse plus large that results in challenges related to performance.
2. Scalability- As users and objects grow; the Collaborative filtering algorithms face scalability issues. Considering tens of millions of objects O(N) and customers O(M), the complexity associated with ‘n’ is extremely large (Koren, Y., & Bell, R., 2015).
3. Synonyms- Objects which are similar but have different names are recognized as different items. The system is incapable of identifying synonymous items.
4. Grey sheep- individuals who their consent do not concur or disagree with the group of individuals can not collaborative filteringfrom CF
5. Privacy, security and trust issues