Temporal filtering system to reduce the risk of spoiling a user's enjoyment

>100 Views

November 01, 20

スライド概要

Temporal filtering system to reduce the risk of spoiling a user's enjoyment

profile-image

Nakamura Laboratory (Meiji University)

@nkmr-lab

スライド一覧

明治大学 総合数理学部 先端メディアサイエンス学科 中村聡史研究室

シェア

埋め込む »CMSなどでJSが使えない場合

各ページのテキスト
1.

Anti-Spoiler: Temporal filtering system to reduce the risk of spoiling a user‘s enjoyment Satoshi Nakamura (Kyoto Univ.) nakamura@dl.kuis.kyoto-u.ac.jp

2.

Backgrounds • There are many content filtering works for harmful contents such as virus, spam, adult content, ... Î They come short of daily usages – People want to filter not only for harmful contents but also for spoilers which spoil user’s enjoyment • The final score of football match • The name of a criminal person in the mystery novel or movie – Huge varieties of spoiler’s sources • News, Weblogs, review sites, Web forums, TV news, radio programs, electric bulletin boards, conversation with friends, ... • New researches increase the risk of spoiling – Recommendation service wants to provide the final score of football match for the user if he/she loves football

3.

Objective • We satisfy the user’s requirement as follows – I want to see it later, but I don’t want to see it now!! • We realize the Anti-Spoiler system – Anti-Spoiler system filters the content which includes the results which spoil the user’s enjoyment partially and temporally

4.

Basic Idea • The system starts running the filter for the target content when the user reserves/buys it • The system clears the filter for the target content when the user finishes enjoying it System Detect (Start) System Detect (End) Filtering for A Reserve A Detect (Start) Filtering for B t Watching A Finish A Detect (End) t Reading B Buy B Finish B User User (A) TV program (B) Book Target content is the content that the user wants to enjoy (recorded TV programs, books, …) Related content is the content which includes result of the target content (news, reviews, …)

5.

Time-based filtering • The system uses the start time of the target content and creation date of the content – The content which was published before starting the target content doesn’t include the result of it • The news page which was published before starting football match doesn’t includes the final score of football match

6.

Requirements • User detection: The system has to detect the user who is in the vicinity of the system. • Schedule detection: The system has to detect the user’s favorite choices, casual schedules, and plans, such as waiting to watch a recorded football game on TV, or waiting to read a recently purchased book. • Activity detection: The system has to detect a user’s current activity, such as watching a recorded program, having finished watching it, reading the book, and having read it. • Filter creation: The system has to create a filter automatically and has to set/unset the filter with the content browser. • Content division: The system has to divide text into several parts, such as topic name (which are itemized in a news site), paragraphs of the text, and weblog entries. • Filtering: The system has to detect undesirable parts in the content by using the preset filter. • Visualization: The system has to prevent undesirable parts from being displayed to users.

7.

Implementation (1/2) • User detection: Using RFID system to detect the user • Schedule detection: – We implement the application to record the TV program – The system detects the book information that the user bought by analyzing e-mail from Amazon.com automatically • Activity detection: – We prepare the application to manage the enjoyment. The user has to check the enjoyed content

8.

Implementation (2/2) • Filter Creation: – The system creates the filter by DB and date • We prepare the sport clubs/players/coaches DB, the keyword DB which has relation to win-loss and the format of score • The system automatically obtains the information of books/movies from Amazon.com • Filtering: – The system runs filtering for the content if it’s last-updated date is newer than the start date of the target content • Visualization: – The system hides part of content based on the risk of spoiling user’s pleasure Filtering Sample

9.

Discussion • We used our system during the 2006 FIFA world cup – The system prevented all results from being displayed – The system sometimes prevented from being displayed the part of the Web content not related to the target contents because some Web contents had no updated information – It is difficult for our system to filter the results of tennis, golf, track and field competition, … – There were some troubles if the user forgot to watch the recorded TV program over the long term – The user has to set the start time if the target content is recorded program • In the future, we plan to … – apply our system to other types information sources, such as TV, radio, advertisement display, word of mouth … – improve the effectiveness of our system – do the evaluation tests of