During the first year, we developed the online argumentation system – iArgue and collected some pilot study data. The platform iArgue allows participants to perform search, discuss, exchange ideas and bookmark relevant information. We also collected 58 college students’ responses for their epistemic beliefs. The findings revealed that students’ epistemic beliefs about “sources” were negatively related to their data used for argumentation. The “development” epistemic beliefs were also negatively related to their claims in argumentation activity. However, “justification” was positively related to their “backing” in the activity. Based on Toulmin’s Argument Pattern (TAP), this study further compared the students’ argumentation behaviors between those in affirmative perspective and those in negation perspective. It was found that there were differences for bookmarking behaviors, searching/storing behaviors and searching titles behaviors. Those in the negation perspective had more frequency than those in the affirmative perspective in these behaviors. During the second year, a total of 145 college students were recruited to participate through iArgue. The participants were divided into two experimental groups (N=66) and two control groups (N=79). The surveys used were Aesthetic Judgment Style Scale (AJSS) and Internet-Specific Epistemic Beliefs (ISEB). The results revealed significant growth in all AJSS factors except for the symbolist judgement style for the experimental groups. There is no significant growth in all AJSS factors for the control groups. With the Add-On budget during the third year, we were able to invited Dr. Mei-Hwa Chen from University at Albany – SUNY to join us to incorporate Knowledge Forum as our online discussion platform for our research. Ninety-eight college students who enrolled in the medical general education course, “Disease medicine and modern” were invited to participate in online discussion through Knowledge Forum. The students were divided into groups to answer open-ended questions, all discussions were recorded on knowledge forum platform. Survey instruments were Medical Epistemic Beliefs Survey and Knowledge Building Environment Scale. Using the factors (multisource, uncertainty, development, justification) in MEB as predictors and the dimensions (working with ideas, assuming agency, fostering community) in KBES as dependent variables, the regression analysis showed that uncertainty was the most significant factor and it can predict all dimensions in KBES.