Design of Recommendation System to Help Undergraduates Select a Course

A summary of group discussion (Josh, Ji Hyun and me) yesterday evening.


Some important features of our proposed online course recommendation system include:


Purpose:
Students need to specify the reason they would like to take a course, is it for future career, personal interest, having fun, looking for different experience, seeking any course available to meet credit requirement, or just to fit into one's class schedule. For different purposes, there would be different focus on the recommendation information. For example, if you like to take the class with your friends, you can find who else already select the course, and other courses those friends are enrolled.

Course evaluation:
Users (undergraduate students) have full access to read the course evaluations /reviews, stat analysis, etc. completed by previous participants while instructors can only view the overall rating.

Kirkpatrick' s four-level framework could be employed to develop the evaluation. At level one (reaction) students assess whether they like the course/instructor or not (something similar as the current evaluation forms). At level two (learning), students may review what they have learned, if the individual learning goals being accomplished, course workload and grading, etc. At level three (transfer) they may comment on how they could use or have applied what they learn, or how it relate to other subjects of their interest. At level four (result), they may find information on how the course may help them qualify for certain jobs or advance in a career.

On the Likert scale, instead of "poor, average, good and excellent", we might use -5-to-5 or 1-10 to obtain more accurate and meaningful assessment. Moreover, if reviewers gave scores extremely low or high, they need to provide specific information to back up their points.

Since reviewers must use their university-assigned user names and passwords to log in the system, it is less likely that they will use fake IDs to boost reputation, or copy-and-paste the same review comments across courses and instructors.

Social components:
On each course information page, there would be links to social networking sites such as Facebook. There will be email links to the particular course instructor and previous participants. Students would be able to see who's online at the same time, and send out IM invitations.

Reputation and Trust

David, S. & Pinch, T. (2006). Six degrees of reputation: the use and abuse of online review and recommendation systems. ?

"Lake Wobegon" appeared at the very beginning of Introduction of which I had no idea. According to Wikipedia, it is a fable place in Minnesota, and a play of words "woe-be-gone," or no trouble at all. The explanation of the Lake Wobegon effect is more interesting: "The characterization of the fictional location, where 'all the women are strong, all the men are good looking, and all the children are above average,' has been used to describe a real and pervasive human tendency to overestimate one’s achievements and capabilities in relation to others."

So online reviews of products (CDs and books in the article) are often inaccurate and conspicuous, yet many people rely on them to assess products. It is not an easy decision to make when offline, it becomes even more complicated online where everybody can post (or copy and paste) a review for various purposes. Further, when reviewers' reputation can be managed through different strategies, trust becomes problematic.

It is critical for novice online consumers to know how the reviewing system works and how to interpret those recommendations and ratings. The authors suggested that the current online reviewing system embedded six degrees of reputation (p.5) is "nuanced" since online behavior is more regulated by code and norms, and less by laws. My question is how the prevalent review system could be improved to break the "cultural Lake Wobegon"? What are the incentives for reviewers to provide more accurate, less-biased recommendations?


Cyr, D., Hassanein, K., Head, M. & Ivanov, A. (?). The role of social presence in establishing loyalty in e-service enviroment. ?

This looks like a neat research paper since it not lonly proposed a model of e-loyalty (elements contributing to the e-loyalty include trust, technology acceptance, enjoyment and perceived social presentce), but also conducted a well-designed experiment to test the hypotheses. Social presense is measured at 5 levels: 1) text and logo, 2) text, logo and photo, 3) live chat added, 4) reading and write review added, 5) all features available. As a result, level 5 of course scored the highest in perceived social presence. It is one thing to claim that increased opportuntities of interaction will lead to better social presence, but another to design and complete a "scientific" experiment to test it.

A question not quite related to the article: is there such a thing as e-loyalty with so many similar websites providing so many similar products and services? I can understand that there might be some sort of attachment to an online community, but I doubt how many online shoppers will be loyal to one or two e-commerce site?

Relationship on the Internet

McKenna, Green and Gleason (2002) suggested that the ability to show "true" self online, the likelihood of close relationship formation, and the possibility of bringing relationships online to offline are mutually related. I would argue that this type of enhancement only exists when a person display a more consistent " true self" online and offline, or there is less discrepancy between the two self's. If a user is intentionally trying to turn into a different person in virtual world, (s)he would keep that online relationship online to be safe.

My early experience/perceptions of how powerful a social network could be was not from BBS, newsgroup, or chat room, but an online game called "Legend". Back in 2002 before my coming to the U.S, it was perhaps the top one game then (still one of the most popular now) in China. Millions of people got addicted to it, including my best friend S. She lived a double life literally. During the day, she's a company employee, well-dressed and self-disciplined. At night, she's a spicy witch and warrior online: besides enemies, she also had her "sisters and brothers", leaders, followers, even pursuers. Her relationships online seemed to be closer and stronger than those developed in her real social network, yet she kept the two worlds/lives apart. Another young couple I knew went even far. The husband met a "girl" in Legend and developed such a "deep" feelings for each other that they eventually got married in the virtual world-- all the time the real wife knew and permitted. When the husband and his virtual bride were holding a wedding ceremony with hundreds of guests and friends attending on the screen, the real wife is sitting right beside him! They actually used the game to escape the boredom of routine life and satisfy their fantasies. However, it is hard to imagine that they would bring the relationships formed online to real life--just too complicated to handle.


Kraut et al (1998) proposed two opposing models--"social enhancement (richer get richer)" and "social compensation"--to predict the relationship between Internet use and social network. I suppose both coexist in reality and insisting on either one tends to ignore the complicity of the problem.
The figure on p.61 shows that as introvert users' use of Internet increases, their community involvement decreases while their feeling of loneliness grows. So if you are lonely offline, going online won't help; or make you feel even worse. Linking back to the topic of building a learning community, the facilitator/moderator/instructor may need pay more attention to this type of learners, making them feel more comfortable, and more confident to participate.

Community and Networks

"A day without Facebook"

Q: So what's the result of boycott? Given the fact that the Feed feature is still in place, it looks that users lost the war. As chimerical company, Facebook of course endeavors to maximize its profit (i.e., by adding Feeds and Ads clicks). Options for users: 1) stop using it if privacy is a priority, 2) change it to add your own protection (by changing the programming?), 3) live with it until you find a new SNS that can do better in balancing privacy and publicity.

"Informal Learning in an Online Community of Practice"

Moderator palys an"absolutely critical" part in "starting up, supporting and sustain" the online learing, formal or informal. "An online moderator combines the roles of technical trouble-shooter, educator, hostess, chairperson, facilitator, and community organizer p.30." --Couldn't agree more.

The case reminded me of my experience of online collaboration study last semester. When the participation was "volunteer-based" rather than mandatory, the biggest challenge became how to keep participants stay. I also used "behind-the-scene" strategies such as sending private emails to thank for postings on line, inviting or encouraging comments, and frequently reminding what to do and what's due. Yet overall they didn't work very effectively, maybe because I'm not as experienced as Gray. :(

Q: When a moderator is a totally stranger to the distance learners, what are some strategies to quickly build rapport, even trust with participants? This could be a challenge in f2f context, too. Some people just seem to be born with the capability to mingle with strangers easily, while others like me feel so difficult to take the initiative in social occasions...again, nature or nurture? culture difference? how to adapt?