In this paper we propose egocentric search methods based on the concept of ”Information and Communicate Activities Navigation (ICAN)” and an authoring support system for Weblog (blog). ICAN regulates the human activities from a viewpoint of information and communication support. We introduce the idea of ”Collect” and ”Relate” in the ICAN table into the information retrieval and the search method which uses contents and human relationship produced by daily blogging. Our egocentric methods provide more subjective search result than the conventional engines. We apply the methods to improve the quality of the small contents made with Weblog tools.
1.1
Computers and networks enrich and facilitate our life so that they now become indispensable for our life. They sometimes enhance our traditional daily activities with their increasing computing and networking power like documenting and communicating with other people, and sometime offer new ways for our activities with new technologies like WWW.
On the other hand, most people become to live with worry that unceasing improvement of computers and networks and installation of new software technologies would change their life and business.
It is not because of such technologies themselves but because of our vision to technologies. We are so eager to develop new technologies that we almost loose the original Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. To copy otherwise, to republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee.
K-CAP2003 Knowledge Markup & Semantic Annotation Workshop ’03
Sanibel Island, Florida USA
Copyright 200X ACM X-XXXXX-XX-X/XX/XX ...$5.00.
mission for development of technologies, i.e., technologies
just for us. Shneiderman pointed out that we should shift
our vision from ”old computing” to ”new computing”. He
explained it in his recent book as follows[
We should shift our focus from information and communication technologies (ICT) to information and communication activities (ICA). We should investigate what are human activities on information and communication and how we can assist people in these activities. 1.2
Human activities on information such as collecting information and communication such as contacting to people are only a part of human activities but they become to play an important role more and more in modern life.
They include various kinds of activities. Shneiderman
shows a simple and therefore understandable model called
ART (Activities and Relationships Table) for them[
To explicate the difference, we propose two-layered model as an extension of his model shown in Fig.1. The first layer has three elements that concern information handling, i.e., Collect (information) Create (information) and Donate (information).
It shows user-centered view of life cycle of information. Information is collected, then new information is created based on the collected information, and finally created information is donated to the society for future creation. It should be noted that new information is seldom created from scratch but created based on existing information.1
The second layer has also three elements that concerns communication handling, i.e., Relate (people) Collaborate (with people) and Present (people).
It is communication-centered view of the above process. People establish relationship with other people, then collaborate with them to create new information, and finally present themselves as donor of new information. Having both information and communication layers is not redundant. What we refer as ”information” in the context of computer technologies is stored data in computers, while human is the source of ”information” in the broader sense, i.e., human can offer information dynamically. We should consider communication in order to include the function ”human as information source”. This parallel view of information and communication activities has thus six categories as activities. Ideally all categories should be supported by computers. Some categories like Collect is well investigated, but others are not. In particular, the three categories in the communication layer should be investigated more.
We aim to investigate information and communication activities and support people in the all categories of the activities. We call such support ”information and communication activity navigation (ICAN)”. It helps people to create new information by guiding information space and human network.
Recently Weblog (blog) or blogging has come into the
spotlight in the World Wide Web[
Some Weblog sites attract the attention with their own editorial policy. The authors of Weblog sites reedit the existing web contents by quoting them. Moreover there are new types of Weblogs that criticize the other Weblogs so that these Weblogs are regarded to organize the ”Weblog community”. There are more than 100,000 Weblogs in the United States Weblogs make people to change from information receiver into information sender and distributor.
Most of Weblog site uses the contents management system (CMS) called Weblog tool. Weblog tools enable the author to describe and edit the small contents via a web browser and transform the contents form text format to HTML files. These tools are implemented based on MVC (Model / View / Controller) model which is the fundamental concept of web applications. The author defines a view template once then do not have to decorate the contents with various HTML tags. This model decreases the cost of publication remarkably comparing with traditional style which requires local text editor and FTP. This feature contributes abundant production of the small contents. Fig.2 shows typical site with Weblog tool.
A huge number of the small contents and citations among Weblog communities are increasing day by day. Some efforts such as topic discovery, trend analysis and content ranking are applied to these large amount of information.
Weblog facilitates to publish the small contents, however, the cost of contents arrangement and classification remains extremely high. Most of Weblog tools are specialized to enhance the convenience of publishing by transforming from text to normal HTML files. There are already billions of HTML files on the Internet so that people are facing troubles both to discover her/his objective articles and to use information effectively. Therefore it is skeptically considered that Weblog will just accelerate this trend. 2.2
There are great hopes that the Semantic Web
technologies will resolve our current condition of information
overload. According to the manifest[
However it is difficult to produce contents with semantic
tags because of their complicated syntax and vocabulary.
Ordinary people hardly find a merit of semantic annotation
because it is a time-consuming task. It is also impossible
to annotate the semantic tags to existing enormous
information on the Internet. There are some researches about
automatic annotation with AI techniques and natural
language processing[
In this research we aim to integrate both technologies, Weblog and Semantic Web, to achieve the platform which enables to share, reuse and reedit our small contents. We provide a new function to the contents management systems like Weblog tools and allow a semantic annotation to existing contents semiautomatically. Hereby it is possible to apply the effects of the Semantic Web to all of the contents on the Internet. As a result, links in Weblogs are transformed into semantic annotations so that the Weblog contents and the web contents refered by Weblogs can be worked as Semantic Web.
In this paper we propose the egocentric search methods with relational annotation and description support system for Weblogging as a first stop of our project.
As mentioned above Weblog tools contribute to increase the amount of small contents. However these tools do not help improve the quality of contents. There is fear that flood of ”junk” contents sweeps the Internet as a result.
We explore new ways of authoring support for blogging. Most of the contents on Weblogs are closely related to the other contents from different sites. Therefore it is important for Weblog authors to know similar or related contents to currently describing contents.
We illustrate the procedures of authoring support according to the ICAN table discussed before2. We assume the user usually refers and comments on several contents on the Internet in her/his Weblog site. These activities associate the user’s contents with the other contents. We define these activities as ”Collect” of information. We also consider these linkages not only as relationships between each contents but relations between the authors who have these contents. This fact corresponds to ”Relate” in the ICAN table.
Thus the contents and human network are built around the user with these Collect and Relate activities.
In case of describing a new content, that is the ”Create”
activity, the authoring support system will retrieve the
contents and human network close to the new content. The
closeness among the small contents is calculated not as the
score of semantic similarity but as the distance from the
user on her/his network. We call these search methods as
”Egocentric search”3. The user will add a new link onto
2Donate and Present are achieved as the original functions
of Weblog that are one of most excellent functions among
other information publishing tools.
3The word ”egocentric” is borrowed from Social Network
Analysis[
We implemented authoring support system for Weblogs with our proposed method as shown in Fig.3. The system consists of three modules as follows.
• Weblog tool
We use ready-made Weblog tool as an infrastructure
of our system. In a number of Weblog tools released
recently we introduce Movable Type system[
We developed an Editor interface as Web application. It can connect the Weblog tools and the Cache Database and execute the egocentric search methods. We will explain it in detail in following section. • Cache Database (DB)
Cache DB stores all contents which is linked in the user’s Weblog. network and the latter is focused in network of individuals. Search engines like google shares the same view with the former, and our approach with the latter. 4.2
Most of Weblog tools generate a RSS (RDF Site
Summary4)[
The RSS, which is the unified regular description format for Web sites, is now propagating from Weblogs to enterprise sites so that people can incorporate various contents into her/his Weblog using the RSS called ”Content Syndication”. However the RSS is for describing content relation in single site, not for inter-site relationship. Therefore we extend the concept of the RSS to describe metadata like inter-site relation. We call this metadata ”RDF Content Summary (RCS)” that is annotated to every entry in Weblog. The RCS uses following modules in addition to the elements of the RSS.
• <semblog:outlink>
<semblog:outlink> module represents an ordinary
hyperlink. ”semblog” indicates the XML namespace we
originally defined. Instance URI of the element is
extracted from <a href> tag in a HTML document.
• <semblog:inlink>
<semblog:inlink> module shows the URI which is
provided as a reverse link (also called ”TrackBack”[
As just described, the RCS maintains both link information of related contents and metadata of entry itself. Furthermore the Movable Type system can generate equivalent RCSs for each entry simply with template. Fig.4 shows an example of RCS file. 4.3
In this section we explain the search methods in the circumstance described previously.
The users daily write and post the small contents to their Weblog sites with the Editor program. The Editor scans the text strings of these contents each times they are posted. If content contains a hyperlink, the Editor acquires whole content and RCS of the link and store it in the Cache DB. Then the Editor extracts hyperlinks from stored contents and constructs a entry network around the user’s contents 4RSS is also a acronym of ”Rich Site Summary” or ”Really Simple Syndication”. with these link information. This network indicates not only relations of contents but also human relationships because all entries on Weblog are owned by an author. Each path of the human networks is weighted relatively to the frequency of citation.
Once the user cites some site as a topic in the new content (entry), the Editor program performs three types of egocentric search and shows the result.
• Relative Chain Search
Relative chain search returns the contents which is directly linked with the entry cited by the authoring content. This model is based on a simple model but consequently it seems most trustful. (Fig.5(a)) • Relative Co-citation Search
Relative co-citation search discovers the entries that link same contents as the authoring entry links to. Co-citation entries are retrieved from the Cache DB and the search result contains the weight of authors. (Fig.5(b)) • Relative Keyword Search
Relative keyword search picks up the entries by keyword matching from the Cache DB. Different from the conventional search engines, our method targets only related sites around the user’s Weblog. (Fig.5(c)) The user read search results by these methods and can append the link of some helpful contents to describing contents. This process may enrich the user’s content and change the search result of the system.
In this paper we propose egocentric search methods based on the concept of ”Information and Communicate Activities Navigation (ICAN)” and an authoring support system for Weblog (blog). ICAN regulates the human activities from a viewpoint of information and communication support. We introduce the idea of ”Collect” and ”Relate” in the ICAN table into the information retrieval and the search method which uses contents and human relationship produced by daily blogging. Our egocentric methods provide more subjective search result than the conventional engines. We apply the methods to improve the quality of the small contents made with Weblog tools.
We will develop the metadata format and the extensions of RSS moreover to represent the relationships among the contents explicitly. We will also provide an egocentric search method based on a pure P2P model, which does not depend on the cache DB. Future model may create a foothold of the Semantic Web for ”the rest of us”. We will take an experimental proof of our system with large Weblog communities in the near future. 6.