- 學位論文
應用群眾智慧解決諮詢問題之探討:以PTT愛情相關看板為例
Consulting with wisdom of crowds: Take love boards on PTT as an example
摘要
中國有一句俗諺：「三個臭皮匠，勝過一個諸葛亮。」意指三個才能平庸的人，若能同心協力集思廣益，也能提出比諸葛亮還周詳的計策。事實上，像這樣結合群眾智慧，使之能與單一智者(專家)相抗衡、甚至超越智者的例子，從過去到現在不勝枚舉。近來，隨著網際網路的發展，應用群眾智慧解決個人/群體特殊問題的系統，更是如雨後春筍般的產生，並且以不同的形式存在且運轉著。最典型的例子就是Wikipedia，它讓人們依照自己的專長與喜好將所知的知識有系統的寫入網頁中，也就是運用了群眾樂於助人的心理，Wikipedia成功建立一個橫跨各領域的「百科全書」；另外，我們每天所使用的Google搜尋引擎也是如此，它最主要是利用網頁與網頁之間的連結關係、以及使用者的點擊率，來判定當使用者輸入關鍵字時，應該首先回傳哪些網頁連結給使用者。其中，網頁與網頁之間的連結、點擊率，實質上都是群眾智慧的累積，很明顯的這些都是單一或是少數專家無法做到的事。儘管此熱潮已經席捲全世界，然而現存運用群眾智慧所建置的系統，大多仍然是運用人類之認知行為層次居多。雖然已有許多的網路平台(例如，Yahoo! Answer +、PTT實業坊)提供較高層次的群眾智慧發揮的場所，但是，這些平台都式開放式的，網友都是進進出出，效果常常都不如預期。 為此，本研究將首先以愛情問題為例，嘗試去瞭解，當人們面臨迷網、無知，甚至抉擇，如何成功藉由網路上開放式論壇的自願群眾(或是稱為鄉民)，幫助自己找到答案或是做出合適的決定，以期未來可以建置更完善之『以網路群眾智慧為基底之線上即時諮詢系統』。 本研究發現，案主初始自我揭露程度越高、案主後來自我揭露程度越高、案主參與討論互動程度越高、吸引鄉民關注程度越高、鄉民與案主互動程度越高，鄉民主動貢獻程度越高，越能獲得群眾智慧的幫助；SVM的結果則發現，當我們將激情、親密、承諾三個向度合在一起看，只要有其中一個向度改變即視為狀態改變，效果就會不錯。
並列摘要
The Chinese saying goes: “three simple cobblers beats a Zhu Ge-liang.” That means if three mediocre people can manage to put their heads together, they can come up with a more well-thought-out strategy than the military strategic genius Zhu Ge-liang. This combining the intelligence of groups to counter or even surpass the intelligence of a single expert has seen many examples throughout history. Recently, with the development of the Internet, we have witnessed many applications of this principle in various systems that leverage the knowledge of crowds to solve the problems of individuals or groups. The classic example is Wikipedia, a platform which allows people to input their knowledge on the Web according to their own specialty or hobby: by thus leveraging the natural desire of people to share their knowledge, Wikipedia has successfully constructed a multi-discipline encyclopedia. Likewise with the Google search engine which we use every day: mainly by using the link relationships between pages as well as counts of user click-throughs, Google attempts to examine the user's search keywords to decide which pages should be listed as relevant, and in what order. These links between pages and click-through rates are in fact accumulations of the knowledge of the crowd -- clearly not something achievable by a single expert or even a small group of experts. This trend seems to have taken over the computing world; however, systems that thus leverage the knowledge of the crowd are still primarily modeled on perceived human behaviors. Although many Internet platforms such as Yahoo, Answer+, and the PTT system offer areas where crowd knowledge is leveraged at a higher level, the open nature of these platforms leads to only sporadic use by users, and thus often less-than-ideal results. In this work we take romance as an example. We seek to understand what happens when a person is emotionally distraught or is looking for information or needs to make a decision; in particular, how the person successfully finds an answer or makes a good decision with the help of willing participants of online open forums (also called “villagers”). Our hope in this is to build a better realtime question-and-answer system based on the combined intelligence of online groups. We find that the more candid the original poster is initially, the more candid the original poster is later on; the more the original poster interacts in the discussion, the more villagers are attracted to participate; the greater the level of interaction between the original poster and villagers, the more the villagers' active contributions, and the more help is obtained via the intelligence of this group of villagers. Results of SVM modeling indicate good performance when combining the three dimensions of passion, intimacy, and commitment, as long as any one dimension's change is taken as a change in status.