Netlogo wiki systems modeling
The paper discusses a wiki-based network communities multi-agent research model. The cases for the model application include a Russian educational wiki portal letopisi.ru and a wiki platform for drafting the Russian Law on education.
Keywords: Wiki, education, Netlogo, simulation.
A significant interest to research of the the potential of network communities, expanding pools of experts and attracting diverse participants to problem solving has been observed in many countries. A number of new social phenomena, such as wisdom of crowds, crowdsourcing, wikinomics and participatory culture has emerged in recent years. All those phenomena imply attracting broad open communities to decision making and collaboration. The range of crowdsourcing actions may vary from simple forms, such as reuse of content objects, media, links etc. to complex processes, such as creation of new documents, books and standards. Understanding crowdsourcing requires tools for analysis and models for simulating its various aspects. We developed a model with which we can discuss and predict behavior of wiki communities in the process of collaborative document creation. The model is based on the Netlogo programming language.
Wikis are collaborative platforms for text creation. The original idea of hypertext, expressed by V.Bush, D.Engelbart and T.Bernes-Lee treated it as an extension not only of individual, but of collective capabilities. Wiki is a simple and radical instance of a collective hypertext, in which every community member can create and edit pages. Wikis are often referred to as tools for conducting collective activities . Wiki philosophy implies aiming the efforts of the whole group at creating a collective final product. A group of wiki users can elaborate a collective hypertext and not bother with maintaining links. Usually wikis are regarded as encyclopedias consisting of multiple interconnected entries or as a multi-agent network community.
In this paper we refer to wikis as ecological systems which consist of multiple human and programmed agents, following certain rules and an environment of various objects, pages, templates and categories and links between them. All the elements of the wiki environment s can be reused. The transclusion mechanism allows to use wikis as building blocks and construct complicated metabolic chains. Well known wiki examples are collections of entries created by the community and are indicative examples of ecological systems. Our model is based on Jyri Engestrom theory that social networks only work if they are organized around a core social object and a verb that defines how people manipulate that object . Engestrom argues that social networks that succeed are based around objects, not relationships. The objects don't have to be physical, but they do have to be distinct entities. Flickr has photos. YouTube has videos. Delicio.us has bookmarks. Wiki has pages. When we consider pages in wiki and their editors, a bipartite network is a convenient representation: U is the set of editors and V is the set of pages in wiki. The bipartite network formalism is ideal for studying collaboration, because the network structure encodes knowledge about which articles editors have edited together.
Netlogo is an effective tool for modeling biological and social ecosystems and can be successfully used by students and teachers . Still as a powerful language for research it can be aligned with Swarm, Repast and MASON. Netlogo was often used for building multi-agent research, sociological, and organizational psychological models.
Netlogo Model of a Wiki System is based on three simple rules. The system contains only participants, pages and links. Participants can act over pages. All participantsҠ actions over the pages are recorded and used for the wiki system dynamics analysis.
Participants possess the following properties:
- Age and status. Every participant has an age which is equal to 0 at his birth. At every
cycle it increases by 1. He also possesses a ԲetirementԠproperty on achieving which he leaves the system. When the age overpasses retirement, the participant ceases his actions.
- Every participant has a status? which can be a user or an administrator.
- Administrators are painted white, they are larger, they never retire. Every participant
has active? property. If a participant becomes older than the retirement age his active? status is switched to false?
Each participant can be connected to several lists:
- impact - is a list which allows to find all pages edited by the participant.
- readlist - is a list of pages read by the author.
- votelist - is a list of pages for which the participant voted. Each participant can vote
for a page only once. He can not vote for the pages which he has created himself.
- Each participant has the following abilities:
- The ability to read pages. After the agent has read a page, it can link it to other pages,
grow other pages from it and link it to other pages.
- The ability to create new pages. A participant creates his page, writes its ID number to
his impact list and writes his ID to the page history. The new page receives a directed pagelink from the parent page.
- The ability to edit existing pages. A participant changes the ԳizeԠ property of the
page, adds its ID to his impact list and adds his ID to the history of the page edits.
- The ability to link pages. A participant creates a directed link between pages. One
page becomes a parent, the other becomes an offspring.
- The ability to evaluate pages. A participant can evaluate pages, created by other
participants and read by him.
On each step user has choice to read, write, edit, link, erase, vote or protect page.
to gowrite if age > retirement [set active? false ht stop] run one-of shuffle (se n-values reading ["read_page"] n-values writing ["write_page"] n-values editing ["edit_page"] n-values linking ["link_page"] n-values erasing [ "erase_page" ] n-values voting ["vote_page"] n-values protecting ["protect_page"] ) set age age + 1 end
Pages can have their own properties inside the system. History is a list of edits and participants who edited the page. Page links to the page and from the page.
Two types of directed links are maintained in the system: Pagelinks are links between pages, established at the creation of a page and as a result of linking two existing pages.
Uplinks are links between participants and pages. These links are created as a result of the system analysis. For creating those links ԃollaborationԠ button is used. Uplinks may vary in thickness, which depends on the contribution of the participant to editing the page.
Fig. 1 The model user interface
The setup button clears off all the content from the system and creates the initial page, protected from erasing. It is a size 2 white page. The On/Off switch controls the OpenWiki variable.
The system can be closed. In this case all the content is published by the administrator.
The system can be open. In this case new users can come and create new pages. Sliders control abilities of the participants.
- reading is the ability to read pages
- writing is the ability to write pages
- linking is the ability to link pages
- editing is the ability to edit pages
- voting is the ability to vote for and against pages
- erasing - is the ability to delete pages
- protecting - is the ability to protect
Collaboration button creates a graph, connecting participants and pages which they edited.
to collab_diagram ask pagelinks [hide-link] ask users [ foreach impact [ifelse is-link? uplink who ? [ask uplink who ? [set thickness thickness + 0.05]] [create-uplink-to page ? ] ]] repeat 10 [layout-spring (turtle-set users pages) uplinks 0.2 5 1 ] end
Links between pages are hidden. The thickness of the link between user and page depends on the number of edits that the participant made (Fig.2)
Fig. 2 Links between participants and pages
Distribution of properties between participants, pages and links can be represented on a histogram, located in the right side of the model. This distribution is typical for scale-free networks.
The model helps us to construct and analyze various wiki behavior scenarios depending on the participants qualities.
Models of wiki cases
First example of wiki project from Russia is the Letopisi project, created in 2006. It initially aimed at setting up a hypertext educational encyclopedia, based on MediaWiki. It is open and accessible to all interested participants. The result of the project is open and accessible to all interested learners http://letopisi.ru Its model is based on a set of simple rules:
- Participants have limited reading abilities.
- Participants have limited abilities to link pages.
- Participants have high ability to create new pages.
As a result participants create separate page branches as it is shown in the Fig. 3.
Fig. 3 Separate page branches
The model of the wiki and the graphs were used in the analysis and discussion of the Letopisi.ru action plan. This model and the data allow making a transition from suppositions to predictions based on experiments with a multi-agent community.
In April 2010 we were approached by the Russian Ministry of Education with a request to develop a crowdsourcing platform for evaluating and editing the draft for the new Law on Education. For testing the model we used the chapter "General education" which consists of 11 articles and 71 items. Our goal was to invite Russian experts, teachers and parents to an in-depth analysis of the Law, evaluating and discussing its overall concept and items, soliciting ideas for improving statements of the law and making new suggestions.
The Law on Education is a major discussion issue in the Russian political life. The quality of education in Russia is one of the major concerns as the vast majority of Russian families have K-12 students. Most discussed issues, such as mandatory graduation testing, religious education and deteriorating quality of public school education are highly debated. We assumed that the community for discussing the subject is very broad and set no limitations on age, affiliation or status of participants.
Our approach to lawmaking is quite different from the existing practice, in which access to creating and making amendments to laws is granted only to a very limited group of "experts", whose expertise is mainly based only to their proximity to the governing bodies. This corresponds with previous research which showed that maximum results are received in crowdsourcing campaigns where certain level of the participants' knowledge of the field is combined with their diversity.
To start the collaborative evaluation/editing process we developed a solution hosted on the Web at edu.crowdexpert.ru. As the front end application we used our own implementation on the base of our own implementation of open source MindTouch wiki with additional deki-scripts. Each item of the Law was posted on the site as a separate page. Upon registration users could make a vote in favor or against the draft, comment on it or suggest their own versions, which was done as a result of saving a previously edited version of an item under the user's name. User created item versions were also commented and evaluated.
Participants use the law items as construction blocks, that can be discussed, evaluated and changed. Every item can be voted for and against and commented on. If the participant thinks that the item should be improved, he can create his own version of the item. The community consisted of 529 participants. 256 of them became co-authors of the final document.
Participants made 2084 votes, 1042 comments and created 95 own versions. 23 of them graduated and were included into the final draft of the document. The initial step of the wiki system model is based on a set of simple rules: The system is populated only with administrators. Administrators publish content for further discussion and editing (Fig. 4).
The next step of the wiki system model is based on a set of simple rules: Participants have high reading ability. Participants can create separate versions of initial (white) pages. On the next step participants vote for initial and user created versions. Versions that receive more votes are selected by the community. Pages receiving more positive votes grow in size. Pages receiving more negative votes grow less. In the end the final draft of the document is put together from positively evaluated pages (Figure 5).
Fig. 5 Selection of approved pages
The model can be used for various wiki scenarios analysis and prediction regardless of the participantsҠ abilities. With the help of the model we observe a shift from a culture of commenting to a culture of collaboration. The 1.0 culture of forums is being replaced with a wiki culture in which participants edit, improve and assimilate content.
- West, James A, (2009). Using Wikis for Online Collaboration: The Power of the Read-Write Web 1st ed., San Francisco, CA: Jossey-Bass.
- Engestrom J. Why some social network services work and others donӴ ֠ Or: the case for object-centered sociality, April 13th, 2005,URL: http://www.zengestrom.com/blog/2005/04/why-some-social-network-services-work-and- others-dont-or-the-case-for-object-centered-sociality.html
- Wilensky, U. (1999). NetLogo. http://ccl.northwestern.edu/netlogo/ Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL.