360 likes | 461 Views
Development of a TV Reception Navigation System Personalized with Viewing Habits. IEEE Transactions on Consumer Electronics, Vol. 51, No. 2, MAY 2005. Tadashi Isobe, Masao Fujiwara, Hiroyuki Kaneta, Toshiya Morita, and Noriyoshi Uratani. bearhsu 2005/11/17. Index. Introduction
E N D
Development of a TV Reception Navigation System Personalized with Viewing Habits IEEE Transactions on Consumer Electronics, Vol. 51, No. 2, MAY 2005 Tadashi Isobe, Masao Fujiwara, Hiroyuki Kaneta, Toshiya Morita, and Noriyoshi Uratani bearhsu 2005/11/17
Index • Introduction • TV Watching Situation Today • TV Navigation System • Evaluation Tests • Results & Assessment • Conclusion
Introduction • motivation • Investigate viewers’ selection behavior • Find out a good selection assisting method • Propose a navigation system for viewers • according to their habits • & occasional feelings
4 Main Facilities • Channel presetting • Program Recommendation • Program sorting • Program retrieval
Index • Introduction • TV Watching Situation Today • TV Navigation System • Evaluation Tests • Results & Assessment • Conclusion
Some Statistics • Since 1974, people watch TV 3.5hrs/day on avg. • Recently increase to 4:05/day • Purposely chosen • Diversion viewing • TV as an environmental furniture
For a viewing individual • Favorite programs • So separate among individuals • Hardly explained by sex, age, occupation… • TV navigation system • Detects viewers’ taste from history • Handles not only concentrated viewing • Casual & diverted ways but also
Index • Introduction • TV Watching Situation Today • TV Navigation System • Evaluation Tests • Results & Assessment • Conclusion
Program selection methods Concentrated viewing Viewing while doing Something else Diversion viewing Doing something With TV on
System Configuration Graphic Display Controller
Service id D, t, u CH(service id) CPS Table (Channel preset Table) Channel Presetting • “Usual one please” D:holiday/weekdays t:time u:user id
Program Recommendation • “I’ll leave it to you” • m= r i e • F= r1 r2 r3 … r256 i1 i2 i3 … i256 e1 e2 e3 … e256 • g=(g1 g2 g3 … g256) • Genre specifying vector
Program Recommendation (cont)’ • RS(t)=mT hu(t) • Recommendation Score • hu(t)= Ru(t) Iu(t) Eu(t) Present Programs, g m=Fg RS(t)=mT hu(t) Top 7 programs at RS
Program sorting • Viewers divided into 8 groups • “Laughter-/stimulation-seeker” • “Diversion seeking zapper” • “Romance-/fiction-oriented” • “Trendconscious TV devotee” • “Easy going interest-seeker” • “Barely interested” • “Wholesome and practical type” • “News-/culture-oriented”.
Program sorting • IS=Gg • Interest Score • G=(G1u G2u G3u… G256u) IS=Gg
Program retrieval • Data in SI words • List up keywords • in alphabetical order • Viewer chooses 1 keyword from list • Enter retrieval algorithm • list appropriate programs • most currently broadcasted first • Viewer can also use a keyword registered in PF
Program retrieval (cont’) Present programs Segment into words Proposed keywords Program retrieval
Emphasis on personalization • Recommendation • A low RS program chosen hu(t) updated • Program sorting • IGF table updated • Once the IGF table is changed, it’s more personalized to viewer • Retrieval • Update both hu(t) & IGF table
Index • Introduction • TV Watching Situation Today • TV Navigation System • Evaluation Tests • Results & Assessment • Conclusion
Program data & Evaluators • Tokyo, 13 days in 2004 • 1 Navigation system/ 1 Evaluator • 38 evaluators in total Older, avg.64 Younger, avg.29
Some discoveries • Mainly 2 kinds of viewing types • News-/culture-oriented • Diversion-seeking zapper • In the morning, Information’s needed
Test procedures/ result • General evaluation • Program recommendation • Program sorting • Evaluators’ impression
General evaluation morning afternoon evening
Some discoveries • “Channel presetting” • is more needed in the morning • “Retrieve” • Afternoon / evening • “Recommendation” & “Sorting” • On a medium necessity
Some discoveries • Morning • P better than A • Afternoon & evening • Older • P better than A • Younger • No significant difference
Some discoveries • Personalizing effect • Still make some effect • But less than “Recommendation”
Efficient vs. Rich/Enjoyable • R: Recommend • A: all 4 • S: Sorting • Y: younger • O: older
Index • Introduction • TV Watching Situation Today • TV Navigation System • Evaluation Tests • Results & Assessment • Conclusion
Results & Assessment • None of the methods was unnecessary • “Channel presetting” => morning • “Program retrieval” => evening • “informative” is needed more • In the morning • Older => efficient • Younger => efficient & Rich/Enjoyable
Index • Introduction • TV Watching Situation Today • TV Navigation System • Evaluation Tests • Results & Assessment • Conclusion
conclusion • Developed a TV reception navigation system easy to use for all generations of viewers • Cope with diversion viewing NOWADAYS • Incorporates recommending facility • None of the 4 facilities is unimportant • Further more… • Optimization of these algorithms would be a more important task