尊敬的
微信汇率：1円 ≈ 0.046166 元
支付宝汇率：1円 ≈ 0.046257元
[退出登录]
雅虎竞拍
煤炉代买
paypay商城
乐天二手
日本亚马逊
乐天新品
ZOZOTOWN
This document summarizes an academic paper on developing an automatic filtering system for online social networks using content-based approaches. It describes a three-tier architecture for the filtering system, with the lowest layer managing social networks, a middle layer performing message categorization and blacklisting, and a top layer providing a graphical user interface. The system works by intercepting messages, extracting metadata using machine learning classification, applying filtering and blacklisting rules, and publishing approved messages while filtering unwanted ones based on content and creator. It aims to allow users more control over messages on their walls by blocking offensive, political, or other undesirable content in an automatic way.
![Proceedings of the International Conference on Emerging Trends in Engineering and Management (ICETEM14)
30 – 31, December 2014, Ernakulam, India
95
2. LITERATURE REVIEW AND RELATED WORK
2.1. Content-based filtering
In content based filtering to check the user’s interest and previous activity as well as item uses by users best
match is found [1].
For example OSNs such as Facebook, orkut used content based filtering policy. In that by checking
users profile attributes like education, work area, hobbies etc. suggested friend request may send. The main purpose of
content based filtering, the system is able to learn from user’s actions related to a particular content source and use them
for other content types.
2.2. Collaborative filtering
Collaborative filtering system selects information item based on user’s preferences, actions and predicts what
users will like based on his similarities to other users. Items are rated on the basis of user likes and dislikes [2].
When
filtering information using Collaborative filtering it contains collaboration of multiple agents. Collaborative filtering
system uses large dataset.Amazon.com uses item to item collaborative filtering for its Recommendation system. The
collaborative approach is suitable for popular items. The content based approach which is more suitable for unpopular
items and effective content information is easily available .The problem of collaborative filtering is to predict how well a
user will like an item that he has not rated given a set of historical preference judgments for a community of users.
C. Policy-based personalization
Policy based personalization is applicable in many different contexts. It adapts a service in specific context
according user defined policies. In online social networking sites user oriented policies can define how communication
between two parties or more can be handled. The policy based personalization system in [3]
focuses on Twitter2. Using
this it assigns a category to each tweet and shows the tweet to the user which are of his interest. Policy based
personalization represent the ability of the user to filter messages according to filtering criteria defined by him .
3. ANALYSIS OF PROBLEM
In the existing system there is a very little support to prevent undesired messages posting in the walls in such a
way that in face book users can specify who post messages in their walls and user can block the person who post
unwanted messages in their walls. Here only block the words from text messages. No Content based preference filtering
are supported in the current system. So we can’t able to filter text messages and also in the current system there is no
method to filter text from the images.
So all the contents posted by others are displayed on our walls. No automatic filtering is supported in current
system.
3.1. Disadvantages of Existing System
Today people misuse liberty of speech in social network. So there is a chance of posting or write down
unwanted words which are treated as unacceptable or uncalled in a civilized society. In the existing system there is no
technical approach to avoid this that is it provides very little support to prevent unwanted messages on user walls. No
content-based preferences are supported. Not possible to prevent undesired messages.
4. FILTERED WALL ARCHITECTURE
The architecture in support of OSN is three tier architecture. The lower layer is social network manager. It
provides the basic functionalities such as profile and it maintains the relationship management. The middle layer is
Social network applications here the message categorization is performed and also blacklist mechanism is applied in this
layer to block the users who send bad contents in their messages. The middle layer consists of two blocks content based
filtering and short text classifier. The top layer is graphical user interface, in this layer the users post his message and this
messages passing through the filtering rules to filter the unwanted contents and the users are block who post unwanted
contents on users wall [4].
When a user tries to post a message on other wall, the filtered wall intercept the user details from social graph
and user profile and metadata from messages are extracted. Then short text classifier divide the messages based on their
content based on the filtering rules applied from the message and from image.](data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7)
