TRANSFER LEARNING FOR RESOLVING SPARSITY PROBLEM IN RECOMMENDER SYSTEMS: HUMAN VALUES APPROACH

Abhishek Srivastava

Indian Institute of Management, Ranchi, India

Pradip Kumar Bala

Indian Institute of Management, Ranchi, India

Bipul Kumar

Indian Institute of Management, Ranchi, India

Address for correspondence: Abhishek Srivastava, Fellow Program in Management, Information Systems, Indian Institute of Management, Ranchi, India. E-mail: abhishek.srivastava13fpm@iimranchi.ac.in; Pradip Kumar Bala, Associate Professor, Information Systems, Indian Institute of Management Ranchi, India. E-mail: pkbala@iimranchi.ac.in; Bipul Kumar, Fellow Program in Management, Operation Management and Decision Sciences, Indian Institute of Management Ranchi, India. E-mail: bipul.kumar12fpm@iimranchi.ac.in

Value	Defining Goal
Self-Direction	independent thought and action--choosing, creating, exploring
Stimulation	excitement, novelty, and challenge in life
Hedonism	pleasure or sensuous gratification for oneself
Achievement	personal success through demonstrating competence according to social standards
Power	Social status and prestige, control or dominance over people and resources.
Security	Safety, harmony, and stability of society, of relationships, and of self.
Conformity	restraint of actions, inclinations, and impulses likely to upset or harm others and violate social expectations or norms
Tradition	respect, commitment, and acceptance of the customs and ideas that one's culture or religion provides
Benevolence	Preserving and enhancing the welfare of those with whom one is in frequent personal contact
Universalism	Understanding, appreciation, tolerance, and protection for the welfare of all people and for nature.

Sparsity %	MAE_Traditional Model	MAE Proposed Model	RMSE Traditional Model	RMSE Proposed Model
80	0.628545209	0.636562104	0.840795043	0.848672753
85	0.655992636	0.660842003	0.87568916	0.878902697
90	0.677086864	0.67306645	0.905027626	0.897071612
95	0.735483466	0.712853471	0.990098643	0.959124849
96	0.783864745	0.7366094	1.046836263	0.98989159
97	0.860027455	0.7488405	1.129117705	1.009658098
98	1.03670117	0.866157436	1.305145056	1.136599866
99	1.288322542	0.957406431	1.485669938	1.228992012

Sparsity %	Precision_5 Traditional Model	Precision_5 Proposed Model	Precision_10 Traditional Model	Precision_10 Proposed Model
80	0.664422388	0.685850817	0.691934323	0.689588426
85	0.641703275	0.651968997	0.678463778	0.678131995
90	0.665792008	0.672231255	0.660809255	0.661584248
95	0.649829764	0.654961295	0.580248654	0.583176702
96	0.589832777	0.621553675	0.537240269	0.564035327
97	0.47318599	0.575101616	0.436581701	0.534757716
98	0.256926952	0.480240922	0.251655629	0.458581505
99	0.070166928	0.380415677	0.070533074	0.376151052

Sparsity %	Recall_5 Traditional Model	Recal_5 Proposed Model	Recall_10 Traditional Model	Recal_10 Proposed Model
80	0.015271725	0.017072546	0.062644194	0.062874904
85	0.028859566	0.030933082	0.114712535	0.116107446
90	0.057823992	0.057988841	0.162363683	0.158737002
95	0.129977215	0.132993659	0.229420491	0.228940217
96	0.151499167	0.159022765	0.229844531	0.241282621
97	0.157856626	0.190731354	0.198696597	0.243410572
98	0.105263158	0.189928058	0.117647059	0.21294964
99	0.021122112	0.16039604	0.031023102	0.165346535

Sparsity %	F1Measure_5 Traditional Model	F1Measure_5 Proposed Model	F1Measure_10 Traditional Model	F1Measure_10 Proposed Model
80	0.029857184	0.033315778	0.114887097	0.11524231
85	0.055235027	0.059063843	0.19624464	0.198268104
90	0.10640658	0.106767567	0.260677725	0.256040911
95	0.216625563	0.221093094	0.328827971	0.328801427
96	0.241077361	0.253251774	0.321950683	0.337983084
97	0.236736923	0.286459095	0.273100147	0.334543783
98	0.149341142	0.272203663	0.160337553	0.290842107
99	0.032469916	0.225650318	0.0430925	0.22971579

{\hat{r}}_{u i} = {\bar{r}}_{u} + \frac{\sum_{j S^{κ_{(v; i)}}} w_{u v^{*}} (r_{v i} - {\bar{r}}_{v})}{\sum_{j S^{κ_{(v; i)}}} w_{u v}_{}}

w_{u v} (cosine s i m i l a r i t y) = \frac{\sum_{i = 1}^{n} r_{u i} . r_{v i}}{\sqrt{\sum_{i = 1}^{n} r_{u i}^{2}} \sqrt{\sum_{i = 1}^{n} r_{v i}^{2}}}

w_{u v} (Pearson c o r r e l a t i o n c o e f f i e c i e n t) = \frac{\sum_{i = 1}^{n} (r_{u i} - {\bar{r}}_{u}) (r_{v i} - {\bar{r}}_{v})}{\sqrt{\sum_{i = 1}^{n} {(r_{u i} - {\bar{r}}_{u})}^{2}} \sqrt{\sum_{i = 1}^{n} {(r_{v i} - v)}^{2}}}

w_{u v} (cosine s i m i l a r i t y) = \frac{\sum_{i = 1}^{n} p_{u i} . p_{v i}}{\sqrt{\sum_{i = 1}^{n} p_{u i}^{2}} \sqrt{\sum_{i = 1}^{n} p_{v i}^{2}}}

w_{u v} (Pearson c o r r e l a t i o n c o e f f i e c i e n t) = \frac{\sum_{i = 1}^{n} (p_{u i} - \bar{p_{u}}) (p_{v i} - \bar{p_{v}})}{\sqrt{\sum_{i = 1}^{n} {(p_{u i} - \bar{p_{u}})}^{2}} \sqrt{\sum_{i = 1}^{n} {(p_{v i} - \bar{p_{v}})}^{2}}}

Brasil