Concept of the Recommender System

Posted on | by Laureen
Do you need this or any other assignment done for you from scratch?
We have qualified writers to help you.
We assure you a quality paper that is 100% free from plagiarism and AI.
You can choose either format of your choice ( Apa, Mla, Havard, Chicago, or any other)

NB: We do not resell your papers. Upon ordering, we do an original paper exclusively for you.

NB: All your data is kept safe from the public.

Click Here To Order Now!

  1. Abstract
  2. Introduction
  3. Related Work
  4. Future Prospect
  5. Discussion & Conclusion
  6. Reference List

Abstract

This paper will be discussing different recommender systems and the benefit they provide for individual usage as well as usage for businesses and education purposes. It will describe different approached and emphasise their advantages and disadvantages. The most popular recommender systems will be analysed in this way. As a result, possible ways of improvement the recommender systems and outcomes of their usage will be provided. The paper will be based on the information received with the help of qualitative research.

A literature review will include twenty articles related to the discussed issue. Thus, the paper will serve as a source of authoritative background information related to the recommender systems and as a guideline for their enhancement. Professionals who work with recommender systems, scientists, students and educators will also have an opportunity to utilise this paper when conducting related research. The paper will start with the explicit summary of related works prepared by scientists and other professionals. Then, a range of possible prospects will be outlined; the future methodology that is going to be used to achieve the research plan will be described; different recommender approaches will be compared; and the conclusion will be made.

Introduction

Our world is constantly altering due to the development in various spheres. Still, the most effect seemed to be provided by the appearance of new technologies. The amount of data related to one object grow, and the process of selection turns out to be more and more difficult. Realising such tendencies, professionals developed search tools that streamline the procedure.

Recommender system works along with Web personalization, which provides an opportunity to base all searches on the user’s preferences that are mentioned in the personal profile (Adomavicius & Tuzhilin 2005). They are made to filter all data and find those that are expected to match with one’s interest. Mainly, recommender systems, the “user profile is keyword-based which, to some extent, involves only those items that match specific keywords in the user’s preferences” (Amini, Ibrahim & Othman 2011, p. 1).

As a lot of business applications became connected with the Web, this sphere also started to face various issues related to the overload of information (Ricci 2002). Both computer and social scientists turned out to be involved in modelling this influence in the framework of Web science. The users received an opportunity to edit information, add commentaries, and upload images or video. Recommender systems are claimed to be very useful for offline and online industries (Michalis & Oikonomou 2011).

Starting with the end of the 20th century, this sphere was constantly developing, which provided an opportunity to research and practice on recommender systems. As a result, a wide range of new business opportunities was developed. Individuals and businesses follow the best practices described in the literature that reveals researchers on recommender systems and provides an opportunity to meet their demands. Scientists investigate various approaches of recommender systems, which allows them to get to know tendencies in customers’ economic behaviour. Except for that, it is a great way to understand business process and competition.

This paper will be discussing different recommender systems and the benefits they provide in the framework of individual Web usage, businesses and education. Advantages and disadvantages of various approaches will be identified and explained. Their analysis will give an opportunity to define the ways of system improvement. Information received on the basis of qualitative research (literature review, in particular) will contribute to the current knowledge as the findings will be used to provide recommendations.

As a result, the paper will serve as a source of authoritative background information related to the recommender systems and as a guideline for their enhancement. Professionals who work with recommender systems, scientists, students and educators will also have an opportunity to utilise this paper when conducting related research.

The paper is full of useful information that deals with the recommender systems. It starts with the explicit summary of related works prepared by scientists and other professionals starting with the end of the previous century, which proves that the discussed issue is worth researching. Then, on the basis of the literature review and found limitations, a range of possible prospects is outlined. Except for that, this section includes expected future methodology that is going to be used to achieve the research plan and a chart that compares different recommender approaches. Finally, the paper includes the discussion of the research process and conclusion that reveals what was done.

Recommender systems started to attract scientists’ attention about twenty years ago. Since that time, a lot of professionals investigated them, which allowed to obtain enough information for this study. The paper will be mainly based on two research works: one prepared by Amini, Ibrahim, and Othman (2011), and one by Michalis and Oikonomou (2011). Except for that, the study will refer to the findings of eighteen other articles related to the discussed topic.

The relationship between knowledge and recommender systems was discussed by Amini, Ibrahim, and Othman (2011). The scientists provide a lot of general information that can be applied to any users. They claim that recommender systems make use of individuals’ profiles and apply various filtering techniques, which allows the users to reach the most authoritative and reliable data that refers to the discussed object or event omitting unnecessary information. The researchers underline the value of user profile.

They state that it has a great impact on the outcomes of the recommendation process because it contains of the data that reveals the things one is really interested in and may be willing to obtain. Paying attention to the way recommendations are affected by the usage of different systems (depending on the source of knowledge), they came to the conclusion that it has a critical role in this process. Professionals outlined peculiarities of every system, stated how the data can be received and what their drawbacks are. On the basis of this information, they encouraged to utilise hybrid approach and make use of several systems at the same time to avoid limitations. They analysed several systems and approaches to define potential improvements that can be helpful for professionals, such as the incorporation of new objects.

In their work, Michalis and Oikonomou (2011) discuss the way recommender systems are used in the framework of marketing and also pay much attention to the context of the recommender system. They emphasise that with the rapid development of the Web and its socialisation, the process of online search became more complicated. In order to cope with this issue, the researchers offer to use collaborative filters that affect the information received by the users on the basis of data about people, events and products. Professionals tend to prove that contextual framework is critical for appropriate filtering of preferences.

Moreover, in this way the users are likely to reach information and products of better quality. The researchers discussed three types of recommender systems that differ in the way recommendations are formed. Thus, they discussed content-based, collaborative and hybrid recommendations in the framework of one-to-one and network-based marketing. The way they affect the processes of personalization and adaptation was also discussed. Burke (1999a) also valued collaborative filtering but with the course of time concluded that hybrid approach is more advantageous (Burke 2002).

Peis, Morales-del-Castillo, and Delgado-López (2008) paid attention to the semantic recommender systems as well as previously mentioned researchers. They provided a bibliographic review, classification criteria, and specific problems that can be limited with the help of these systems. Bedi, Kaur, and Marwaha (2007) also discussed semantics but in the framework of trust-based systems. Unlike them, Pazzani and Billsus (2007) paid attention to the content as a basis for the recommendation.

Recommender systems can are taken as the basis for personal shoppers. Burke (1999b) claimed that they can be used as an independent browsing tool. In this way, the information is accessed online and is used to select particular products from the web-based catalogues. The process is maintained with the text search or query formulation. It is a preference-based navigation that is applied for commercial purposes. Sarwar, Karypis, Konstan, and Riedl (2000; 2002) also discussed e-commerce.

Oard and Kim (1998) claim that recommender systems affect the information space. Professionals focused on the possibility of utilisation implicit feedback instead of explicit ratings, paying attention to the fact that they make databases become beneficial for one or several users. All in all, they concluded that observations can be discussed along with the predicted ratings due to their interrelation.

Recommender systems can also be effectively used in the framework of the movie industry (Vozalis & Margaritis 2007). With the help of PCA-Demog algorithm, professionals offer to combine different data sources to make the most accurate predictions. Such approach seems to resemble a filter one due to the utilisation of combined algorithms.

Das and Horst (1998), in their turn, paid attention to the way TV users can benefit due to the usage of recommender systems. They mentioned that both implicit and explicit profiling may be effectively used to develop a rating of programs or films that are likely to appeal to the particular consumer or a group of consumers. Program categories can be created to simplify the search. The same aspect was also discussed by Gutta et al. (2000).

Future Prospect

Efficient and effective outcomes of the recommender systems usage depend greatly on the accurate personalization according to the words of Amini, Ibrahim, and Othman (2011). Researchers offered to extend recommendation frameworks “by incorporating new objects such as deep contextual features, user’s cognitive properties, and case-sensitive mutli-criteria ratings into the recommendation process in order to increase the accuracy and performance of recommender systems” (Amini, Ibrahim & Othman 2011, p. 11). Still, they underline that the concept of context is perceived differently by various professionals, which may cause particular misunderstandings.

Except for that, they mentioned that their approach of recommender systems improvement was proposed for the first time and lacks investigation even though the implementation of semantic approaches was already proved to be advantageous. Thus, it is necessary to develop ideas of these professionals and define the ways in which recommender systems can be used, pointing out peculiarities of their utilisation and ways to overcome limitations outlined by the professionals. Finally, the framework of the paper can be narrowed; for example, the way businesses make use of the systems in order to divine their customers’ interests can be investigated.

Michalis and Oikonomou (2011) supported the view of previously mentioned researchers regarding the value of context and stated that it affects the efficiency of recommender systems. Professionals paid attention to the fact that recommender systems are not going to remain static. Constant technological development and market expansion create new challenges for them, such as issues with privacy, the ubiquity of the environment, and competition.

Thus, it would be necessary to pay attention to both technical and business-oriented peculiarities. Michalis and Oikonomou encourage other researchers to continue their study and investigate these challenges of recommender systems as well as the ways in which they can be overcome. Along with Peis, Morales-del-Castillo, and Delgado-López (2008), they advise to pay attention to the value of semantics.

The further research can also be focused on the work of Gipp, Beel and Hentschel (n.d.). Professionals discussed the usage of the recommender system in the academic framework, as well as Lu (2007) and Kabore (2012), and came up to the conclusion that the developed Scienstein system is extremely advantageous for the students, as it provides an opportunity to take into consideration not only keywords but also “citation, author and source analysis, similar and related documents” (Gipp, Beel & Hentschel, n.d., p. 1).

Still, the study did not discuss non-technical aspects of the system. In this way, the future paper can be focused on privacy concerns. Linton, Charron, and Joy (1998) also discussed how recommender systems can be used for educational purposes but in the organisational environment. Middleton, Roure, and Shadbolt (2002) believe that it is possible to create quality ratings with the help of recommendation systems, which can be proved or disproved in our future research.

In order to achieve mentioned goals, a thorough research will be conducted. All relevant information will be gathered from the literature sources in order to discuss those issues that were not previously explained. As many sources were written more than ten years ago, up-to-date information will be included. As people started to use recommender systems more often due to the technological development, it would also be advantageous to gather their feedback. The paper will utilise information that reveals how these systems are currently used and in what way they are improved.

Recommender Systems Algorithms

Paper title Recommender system name Reason of using recommender system Method used Algorithm used Advantages of recommender system Disadvantages of recommender system
Discovering the impact of knowledge in recommender systems: a comparative study. Knowledge-based recommender system. To get pertinent recommendations from the decision rules. Knowledge-representation methods.
Machine learning.
Knowledge engineer interaction.		 User-based recommendation algorithm. Exactly match user’s interests according to the profile.
Direct interaction between users and the system.		 Subjective and static profile.
Content-based recommender system. To get pertinent recommendations from the content of visited Web pages. Document modelling.
Information filtering.
Information extraction.		 Content filtering algorithm. Recommendations on the basis of the description of previously seen information.
Define an interest degree.		 Overspecialization.
Dependence on the availability of content.
Not considered semantic meaning.
Collaborative-based recommender system. Avoid limitations of content-based system. K-Nearest Neighbour.
Similarity.		 Collaborative filtering algorithm. Consider interests of the whole community. Spare coverage.
Latent state.
New item rating.
New user.
Cold-start.
Violated privacy.
Demographic-based recommender system. Group users according to their demographic characteristic. Classification methods.
Locating group interests.		 Social filtering algorithm. Recommendations for a particular category of people. Dependency on the availability of demographic data of poor quality.
Recommendation systems: a joint analysis of technical aspects with marketing implications. Content-based recommender system. To get pertinent recommendations from the content of visited Web pages. Document modelling.
Information filtering.
Information extraction.		 Content filtering algorithm. Recommendations on the basis of the description of previously seen information.
Define an interest degree.		 Overspecialization.
Dependence on the availability of content.
Not considered semantic meaning.
Collaborative recommender system. Avoid limitations of content-based system. K-Nearest Neighbor.
Similarity.		 Collaborative filtering algorithm. Consider interests of the whole community. Spare coverage.
Latent state.
New item rating.
New user.
Cold-start.
Violated privacy.
Hybrid recommender system. Avoid limitations of content-based and collaborative systems. Content and collaborative methods. Item-based top-N recommendation algorithm. Consider interests of the community on the basis of the information its representatives prefer. Dependence on the availability of content.
Scienstein: a research paper recommender system Scienstein. Advanced academic search. Distance
Similarity Index.
In-text Impact Factor.		 Nearest neighbour.
Bibliographic coupling.
Collaborative filtering.
Further paper.		 New terms associate to existing documents.
Well-structured classifications.
Multiple tags.
Creation of further categories.		 Cannot identify homographs.
Not all papers are in database.
Irrelevant entries.
Unclear nomenclatures.
A recommender system using principal component analysis. The GroupLens movie recommender system. Receive accuracy values.
Get the clearest predictions.		 Principal component analysis. PCA-Demog algorithm. Considers both user ratings and demographic characteristic. Sparsity and the synonymy problems.
Recommender systems for TV. TV Advisor. Value the cost of precision.
Always receive recommendation.		 Elicitation method.
Similarity
Browsing.
Query-by-example.		 Social and collaborative filtering. Confident recommendation.
Multiple designs, implementations and evaluations.
Sensible
Recommendations even if there is no match.		 A lot of user interface requirements.
Unclear domain network connections.
Semantic recommender systems. Analysis of the state of the topic. Semantic recommender system. Work with synonyms, homonyms, similar tags. Similarity.
Ontology Web Language.		 Nearest neighbour-based recommender algorithm. Distinguish data collection systems.
Consider semantic meaning.		 Scalability problem.
New users.
Security issues.

Discussion & Conclusion

This paper was focused on the recommender systems and different approaches to their usage. It summarised information received during the literature review and defined several perspectives for future research. The paper included the comparison of the recommender system approaches that can be utilised to identify the most effective and efficient ones. It focused on individuals, businesses and education, which provided an opportunity to see how these systems can be used and for what purposes. It was stated that the paper can discuss the ways to make recommender systems more efficient and the recommendations they make more valuable. The ways, in which this data can be found, were also mentioned.

When preparing this paper, I realised that today scientists pay less attention to the investigation of the recommender systems. It surprised me, as technologies are constantly developing and alterations are needed. Unfortunately, this paper does not include the way the general public treat recommender systems and what they know about them, but I would like to have such insight. For now, this paper just provides a background and perspectives for future research. Still, when being finished, it would include new information, helpful to both students and professionals.

Reference List

Adomavicius, G & Tuzhilin, A 2005, ‘Towards the next generation of recommender systems: a survey of the state-of-the-art and possible extensions’, IEEE Transactions on Knowledge and Data Engineering, vol. 17, no. 6, pp. 734-749.

Amini, B, Ibrahim, R & Othman M 2011, ‘Discovering the impact of knowledge in recommender systems: a comparative study’, International Journal of Computer Science & Engineering Survey, vol.2, no.3, pp. 1-14.

Bedi, P, Kaur, H & Marwaha, S 2007, Trust based recommender system for the semantic web.

Burke, R 2002, ‘Hybrid recommender systems: survey and experiments’, User Modeling and User-Adapted Interaction, vol.12, no. 4, pp. 331–370.

Burke, R 1999a, Integrating knowledge-based and collaborative-filtering recommender systems.

Burke, R 1999b, .

Das, B & Horst, H 1998, .

Gipp, B, Beel, J & Hentschel, C n.d., .

Gutta, S, Kurapati, K, Lee,K, Martino, J, Milanski, J, Schaffer, D & Zimmerman, O 2000, , 2016.

Kabore, S 2012, .

Linton, F, Charron, A & Joy, D 1998, .

Lu, J 2007, A personalized e-learning material recommender system.

Middleton, S, Roure, D & Shadbolt, N 2002, .

Michalis, V & Oikonomou, M 2011, .

Oard, B & Kim, O 1998, .

Pazzani, M & Billsus, D 2007, Content-based recommendation systems.

Peis, E, Morales-del-Castillo, J & Delgado-López, J 2008, Semantic recommender systems. analysis of the state of the topic.

Ricci, F 2002, Travel recommender systems.

Sarwar, B, Karypis, G, Konstan, J & Riedl, O 2002, Recommender systems for large-scale e-сommerce: scalable neighborhood formation using clustering.

Sarwar, B, Karypis, G, Konstan, J & Riedl, O 2000, .

Vozalis, M & Margaritis, K 2007, .

Do you need this or any other assignment done for you from scratch?
We have qualified writers to help you.
We assure you a quality paper that is 100% free from plagiarism and AI.
You can choose either format of your choice ( Apa, Mla, Havard, Chicago, or any other)

NB: We do not resell your papers. Upon ordering, we do an original paper exclusively for you.

NB: All your data is kept safe from the public.

Click Here To Order Now!