台灣有很多的新創公司，產業結構以中小企業為主， 2013年台灣中小企業佔全體企業的97.64%；不可否認的是中小企業是台灣經濟的中流砥柱，也是幫助就業與平均分配所得的要角。然而，有超過三成的新設企業在五年以內結束營業，而經濟部2013年所出版的全球創業觀察台灣年報中的調查表示，新設企業在資金周轉不靈或是資源缺乏的情況下，將難以持續營運企業。 當人們再購買產品或是服務之前，多半傾向會去詢問親友或是熟人的意見，有93%的消費者會上網搜尋口碑意見，且有近53%的消費者會上網評論、分享他們的經驗，而當對於商品或服務滿意時，評論的意願會增加。國內外對於口碑效應的研究不在少數，但研究本身多集中在口碑效應的行程以及消費者行為，較少從企業的觀點出發研究口碑管理，如果能夠深入研究口碑管理的操作細節並能夠為新創事業所運用的話，將能夠大大提升資源效率與公司的存活率。 本研究欲探討的問題有三，1. 新創事業在達到穩定期之前如何促進口碑傳播？再者，促進口碑傳播的行為與其成長階段的關係為何？2. 新創事業在達到穩定期之前如何監視口碑？再者，監視口碑的行為與其成長階段的關係為何？3. 新創事業在達到穩定期之前如何回應口碑？再者，回應口碑的行為與其成長階段的關係為何？ 透過深入訪談新創事業-木酢家，並配合相關次級資料的蒐集與整理，本研究獲得以下結論1.新創事業會透過接觸虛擬和實體世界中不同的人際網絡促進口碑傳播。2. 新創事業口碑監視的重點在於，更詳細的了解消費者行為，以及顧客真正想要的產品。3. 口碑回應是對消費者傳達理念的途徑，新創事業再塑造形象時會納入策略的一部分。 / There are many start-up companies in Taiwan, and therefore made the industrial structure full of small- and medium-sized enterprises (SMEs). In the year of 2013, SMEs account for 97.64% of all the enterprises,and are the mainstay of the employment market andthe economy system in Taiwan. However, the data shows that over 30% of SMEs end up in 5 years, the Global Entrepreneurship Monitor (GEM) from the Ministry of Economic Affairs said that the lack of money and resources are the biggest challenge of running a start-up company. When making the decision of buying products and services, people incline to ask for the opinions from friends or acquaintances. The data shows that 93% of the respondents search word-of-mouth informations on the internet, and 53% of them shared their experiences. The more satisfaction they perceived, the more likely they would share the experience. There were abundance of researches in the past studying word-of-mouth, much of them gave discussions on word-of-mouth effects and consumer behaviors, while only a few studying about word-of-mouth management from the enterprise view of point. Despite the power that word-of-mouth contains, if it can be apply by some of the SMEs, it might enhance the efficiency of utilizing resources. This thesis is going to study three questions: 1. How does a start-up company foster word-of-mouth before the company comes to a stabilizing phase? Moreover, what is the relationship between the word-of-mouth foster activity and the growing phases? 2. How does a start-up company monitor word-of-mouth before the company comes to a stabilizing phase? Moreover, what is the relationship between the word-of-mouth monitor activity and the growing phases? 3. How does a start-up company responds to word-of-mouth before the company comes to a stabilizing phase? Moreover, what is the relationship between the word-of-mouth responding activity and the growing phases? The research findings are describe as following:1. A Start-up company foster word-of-mouth by attaching different virtual and physical social networks. 2. The word-of-mouth monitor activity of a start-up company focuses on the comprehension of consumer behaviors and the consumer needs. 3. The word-of-mouth responding activity is a way to convey the belief of the company, it would be a part of the strategy when a start-up company is shaping its image.

本研究探討中文讀者對於中文雙字詞的早期語意處理，特別要探討的是中文讀者是否能在凝視一個中文雙字複合詞之前，即可提取該詞的語意資訊。在中文的閱讀研究中曾經發現單字的語意資訊可以在該單字被凝視之前提取，不論是成詞的單字或是多字詞其中的組成單字。也有證據顯示中文雙字詞或雙字複合詞呈現在中央視野時的處理方式是整詞處理。由於事實上就詞長來看，雙字詞的詞類數量為中文詞類的最大宗，也是最常被使用的詞類，因此中文雙字詞語意處理的時間歷程便是本研究的研究目標。 　　實驗一旨在檢視雙字詞的語意資訊是否能和單字一樣，在被凝視之前即被提取。本實驗採用邊界典範（Boundary paradigm, Rayner, 1975），除了目標詞預視（identical preview）之外，本實驗亦包含語意相關預視（semantic-related preview）、語意無關預視（semantic-unrelated preview）及非詞預視（nonword preview）。實驗發現語意相關預視能促進目標詞的處理。然而和預視空間（preview space）及預視時間（preview time）的交互作用則顯示語意預視效益（benefit）在預視空間較大的時候會隨著預視時間拉長而增加，在預視空間較小的時候，則會隨著預視時間漸減。 　　在實驗一中，語意關聯性的高低和該詞合理性（plausibility）的高低是共變的，因此這兩個要素構成一個混淆的因子。語意相關預視和語意無關預視之間的效果有可能來自語意關聯性的差異，或是來自在句中合理性的差異。合理性同時也能解釋在為何在實驗一中，不合理的語意無關預視在目標前詞（pretarget）上造成較短的閱讀時間。為了解決這項混淆的因子，實驗二因此採用對目標詞預視、語意相關預視、及語意無關預視都合理的句子。結果發現，語意預視的主要效果消失。然而交互作用的模式則顯示出，語意預視效益在較長的預視時間、較大的預視空間下仍會存在。但在較小的預視空間下，會隨著較長的預視時間而轉為耗損（cost）。實驗二的結果因此提供證據支持在沒有合理性的交互作用下，中文雙字詞的語意預視效應仍會發生。最後，兩個實驗的差異顯示在中文閱讀中存在合理性的預視效益，此結論和之前中文閱讀的研究結果一致。 / The present study investigates the early semantic processing of Chinese two-character words by Chinese readers. Specifically, whether Chinese readers are able to extract semantic information of an up-coming two-character compound as a whole when the word is yet being fixated. In Chinese, it has been demonstrated that semantic information can be extracted from a single character, whether it is a word or part of a word, before the character is being fixated. There is also evidence for whole word processing of foveally presented two-character compounds/words. Since two-character words actually constitute the majority of word type and are used most frequently in total, the time course of processing the meaning of such combination of characters during reading is then the goal of this study. The first experiment aimed to examine whether semantic information of a two-character word can be extracted before it is fixated, as what have been found for single characters. Boundary paradigm (Rayner, 1975) was used, with identical, semantic-related, and semantic-unrelated words, as well as nonwords as preview. Semantic-related preview did facilitate target word processing. The interaction pattern of the effects with preview space and preview time, however, showed that semantic preview benefit could increase with preview time with small preview space, but decrease with preview time under large preview space. A possible confounding factor in the first experiment was the overlap between semantic relatedness and plausibility. The effect between semantic-related and semantic-unrelated previews could be of semantic or plausibility nature. Plausibility may also explain the shortened fixation duration found in Experiment 1 when implausible semantic-unrelated preview was presented parafoveally. Experiment 2 then solved this confounding by using sentence frames which are plausible for identical, semantic-related, and semantic-unrelated previews. In Experiment 2, main effect of semantic preview benefit disappeared, while the interaction patterns showed that such benefit existed for large preview space with long preview time, but became cost for small preview space with long preview time. The results of Experiment 2 thus provide evidence for semantic preview effect of Chinese two-character words without the interaction with plausibility. Finally, the discrepancies between the two experiments indicate the existence of plausibility preview benefit, which previous studies have suggested to exist in Chinese.

迴歸分析中，若變數間有非線性(nonlinear)的關係，此時我們可以用B-spline線性迴歸，一種無母數的方法，建立模型。Penalized spline是B-spline方法的一種改良，其想法是增加一懲罰項，避免估計函數時出現過度配適的問題。本文中，考慮三種方法:(a) Marginal Mixed Model approach, (b) Conditional Mixed Model approach, (c) 貝氏方法建立信賴帶，其中，我們對第一二種方法內的估計式作了一點調整，另外，懲罰項中的平滑參數也是我們考慮的問題。我們發現平滑參數確實會影響信賴帶，所以我們使用cross-validation來選取平滑參數。在調整的cross-validation下，Marginal Mixed Model的信賴帶估計不平滑的函數效果較好，Conditional Mixed Model的信賴帶估計平滑函數的效果較好，貝氏的信賴帶估計函數效果較差。 / In regression analysis, we can use B-spline to estimate regression function nonparametrically when the regression function is nonlinear. Penalized splines have been proposed to improve the performance of B-splines by including a penalty term to prevent over-fitting. In this article, we compare confidence bands constructed by three estimation methods: (a) Marginal Mixed Model approach, (b) Conditional Mixed Model approach, and (c) Bayesian approach. We modify the first two methods slightly. In addition, the selection of smoothing parameter of penalization is considered. We found that the smoothing parameter affects confidence bands a lot, so we use cross-validation to choose the smoothing parameter. Finally, based on the restricted cross-validation, Marginal Mixed Model performs better for less smooth regression functions, Conditional Mixed Model performs better for smooth regression functions and Bayesian approach performs badly.

自Greenberg (1972) 以來，陸續有學者探討世界語言中的複數標記和分類詞之間的關係，並指出兩者的相似之處(Greenberg, 1972; Sanches and Slobin, 1973; Borer, 2005; Her, 2012)。本文稱這些研究為CL-PM Convergence View。這些研究指出分類詞 (numeral classifier; CL) 與複數標記 (plural marker; PM)呈現互補分布的關係。學者們 (Borer, 2005; Her, 2012) 更進一步認為，分類詞與複數標記在同一名詞組裡呈現互補分布，表示兩者在句法結構上佔據相同的位置，應視為相同的成分。然而本文發現，台灣華語「Num+CL+N們」結構有一定的能產性。若將「們」視為複數標記，則台灣華語對於上述學者的理論就形成了反例。有鑑於此，本文目的在於以句法接受度實驗及語料庫兩項方法，重新檢視台灣華語「們」的各種用法，以期解決文獻上對於「們」的諸多爭議。接著，在建立語言事實後，本文探討「們」對於CL-PM Convergence View的意義，並對於台灣華語「們」與分類詞在歷時和共時上的互動作出新的解釋。 具體來說，本文釐清了下列六項事實，為文獻上的爭議提供新的證據: (一)「們」可用於非指人的名詞；(二)「N們」為定指；(三)「Proper N們」僅表示 ‘多位Proper N’；(四)「¬1群N們」合法；(五)「Num+CL+N們」合法；(六) 在有接受英語教育的前提下，母語者的英語程度越低，對於「Num+CL+N們」的接受度越高，也越容易受到英語句法結構的促發(priming)，表示「Num+CL+N們」的產生與和英語的接觸有密切關係。考慮這些事實及其他台灣華語沒有爭議的特性，本文認為台灣華語「們」應視為一個集合標記(collective plural marker)，而非文獻所說的伴同標記(associative plural, Iljic, 2001,2005; 陳俊光，2009)或是普通複數標記(additive plural, Li, 1999; Hunag et al, 2009)。最後，本文提出兩個論點: 第一，我們根據Her et al (to appear)的洞見，區分「語意複數」(semantic plural)及「語法複數」(grammatical plural)；第二，我們提出新的事實，論證「們」在句法上，台灣華語「們」是一個附綴(clitic)而非詞綴(suffix)。這兩項論點證明台灣華語「們」並不違反CL-PM Convergence View的預測，亦可以解釋「們」與分類詞在歷時(李豔惠、石毓智，2000)和共時上的互動。 / Since Greenberg (1972), there have been many studies addressing the issue of the relationship between numeral classifiers (CL) and plural markers (PM) (Greenberg, 1972; Sanches and Slobin, 1973; Borer, 2005; Her, 2012). These scholars (henceforth CL-PM Convergence View) discovered that CL and PM tend not to co-occur in the same language, and even if they do co-occur, they are complementarily distributed within NP. Some linguists (Borer, 2005; Her, 2012) take this generalization further to propose that CL and PM in fact belong to the same category. However, when we look at Taiwan Mandarin (TM) data, a potential counterevidence can be found: [Num+CL+N-men], in which CL and –men, generally taken to be a plural suffix, co-occur within the same NP. In light of this, this study aims to take a realist look at TM –men, collecting relevant data from grammaticality judgment task and corpora so as to capture the behavior of –men. We then test CL-PM Convergence View against empirical data obtained in the study, showing that [Num+CL+N-men] does not constitute a counterexample to CL-PM Convergence View. The apparent interaction between CL and –men in TM can also be accounted for under our analysis of –men. Specifically, this study establishes the following facts for TM –men: (1) the use of –men is not restricted to human Ns; (2) N-men must be definite; (3) Proper N denotes ‘more than one Proper N’; (4) [¬1 qun N-men] is grammatical; (5) [Num+CL+N-men] is grammatical; (6) native speakers’ acceptability of [Num+CL+N-men] is in negative correlation with their English proficiency, and priming effects of English structure [Num+N-s] are observed on speakers with low English proficiency. Taking these findings into account, this study proposes that TM –men should be best analyzed as a collective plural marker, contra Iljic, (2001,2005) and 陳俊光’s (2009) “associative” analysis on the one hand, and Li (1999) and Hunag et al’s (2009) “additive” analysis on the other. Accordingly, we argue that –men as a collective does not constitute a counterexample to CL-PM Convergence View, citing two further pieces of evidence: Her et al’s (to appear) insight that “semantic plural” and “grammatical plural” should be distinguished and the proposal made there to revise CL-PM Convergence View, and the “clitic” analysis of TM –men proposed in this study. Finally, we show that the distinction between “semantic plural” and “grammatical plural” also nicely explains the synchronic and diachronic interaction between CL and –men in TM.

隨著各種無線網路技術興起至今，各類無線網路技術及產品日益成熟與普及且無線網路的使用者也逐漸變多。而今在各種無線網路的廣泛使用下、時而可見的是當一個行動使用者手持行動裝置身處在一地理環境中，手持行動裝置一經掃描後，常會發現有許多無線網路其覆蓋範圍共同覆蓋在同一個地理區域中，此時多個無線存取點(Access Point)就會發生訊號覆蓋的重疊現象。行動使用者的行動裝置常需判斷是否應該切換到另一個無線網路，此種情形也就是所謂的遞換(Handoff)情形。遞換之產生，代表行動使用者從原本的無線網路區域範圍移動到另一個無線網路範圍，此時若因外界無線電訊號有所變化或是相關網路協定運作出現狀況時、就很容易產生通訊中斷的情形。 在本論文中我們將探討在異質性網路的情況下，根據多種不同的參數、來 設計一個垂直交握(Vertical handoff)的機制。以克服在異質性網路中，時常發生連線中斷之問題。本篇論文主要引使用模糊理論概念來推論其遞換(Handoff)執行的時間點，並且明確的告知終端設備要進入何種網路。 / In this paper, we propose a terminal –control scheme to avoid unnecessary vertical handovers between 3G cellular and WLAN hotspots. The proposed algorithm is a fuzzy logic based algorithm that can adapt itself to the dynamic conditions of the hybrid networks. The scheme can reduce the access delay, unnecessary handoff probability and aid the handoff decision for better connectivity. This handoff algorithm has much better performance than conventional algorithms and can be extended to other heterogeneous networks.

日本為東北亞地區的主要國家，自二次大戰後受限於和平憲法規定，被迫禁止硬權力的發展；為求國力發展，日本以戰後重建、恢復經濟、繁榮富裕的和平原則，成為經濟大國。而日本海外直接投資（FDI）在對外經濟活動中所佔比例極重，不僅對日本經濟發展貢獻良多，在日本對外關係中也扮演重要角色；透過海外的大量投資、專業技術的把持，使得日本在國際關係中扮演舉足輕重的角色，而80年代末東亞地區盛行的雁行模式，便是由日本海外企業投資所主導。 日本在國際關係上，向來為美國最重要也最忠實的伙伴；缺乏自衛武力的日本，在冷戰背景下，受到美國核傘的保護，並於國際事務上追隨美國的全球戰略，並以此原則確立冷戰時期日美外交事務的基軸。在吉田主義的指導原則下，日本致力於經濟復甦，而美國則扶持日本作為其戰略伙伴、大力支持日本經濟發展，並開放美國市場，協助日本產品進入美國。 隨著日本經濟日漸發展，日美貿易產生順差，而後發展為日美貿易摩擦，最後在美國的壓力下，日本於1985年簽訂廣場協議，逼迫日幣升值；受到匯率的影響，日圓不斷升值，在1985年至1996年間，日幣對美圓的比例由250：1遽升為87：1。日幣的升值固然在國際上削弱日本產品的競爭力，但美元的相對貶值，卻造成日本的資本輸出熱潮。從1986年至1991年，日本的海外投資總額達4000億美元，成為全球最大的海外直接投資輸出國。1985年後，日商公司開始大量購買美國企業，或在美國開設工廠，日商公司的瘋狂收購行動，也引起美國民眾的恐慌。而此時期的日本企業因缺乏海外投資經驗，在資訊不明、非理性決策模式下大量收購收購美國資產，導致日商在期貨、地產等行業的損失慘重。 本文將日本海外直接投資的研究範圍設定在1994年至2007年，主要是因為此時期的FDI不同於80年代末期的日本對美投資；不同於以往海外投資失敗的經驗，日本企業將國內失去比較利益的產業外移至其他工資低廉、生產成本較低的開發中國家，為貿易取向的海外投資。而日商企業至歐美等先進國家的投資，則是著眼於當地市場，為市場導向投資。隨著不同時期重大事件的發生，日商企業微調自己在全球佈局的戰略方向：NAFTA的成立，促使日本跨國公司加快前往墨西哥投資的腳步、亞洲金融風暴雖然重創該區開發中國家的經濟，但卻增強亞洲產品的競爭力，鑑於對亞洲開發中國家的經濟成長潛力與未來市場發展，日商企業依舊苦撐在此地區的生產據點、中國低廉的勞工成本則是吸引日商投資，造成熱潮、隨著中國經濟發展，工資相對於越南、印度提高，使得日商企業思考中國以外的投資熱點、由於印度等金磚四國的崛起，日商的投資觸角也延伸至南亞地區。 在全球化的浪潮下，日本的海外投資已不再是單向行為，不僅影響與地主國間的經貿關係，也牽動著第三國的貿易活動：日本對中國的直接投資佔中美貿易順差絕大比例，透過對中國直接投資，將產品出口至美國，日商企業藉著中國的低廉工資降低生產成本，也縮減日美貿易順差，將此缺口部分轉移至中美順差。產品所標示的「中國製造」標籤，只能說明產品的最後製造地為中國，但生產資本投入、中間財與獲利企業可能是來自於不同國家。 2008年自美國掀起之世界金融海嘯，較1997年亞洲金融風暴、2001年IT產業泡沫化更加嚴重，普遍造成全球國家景氣衰退、企業獲利短缺，在一片悲觀的氛圍下，仍有日商企業於海外投資活動中獲利。這些企業與眾不同的地方，正是在金融海嘯下生存的關鍵；日商企業的研發能力與不斷求新、求進步的活力，更是台灣企業值得借鏡的地方。 關鍵詞：海外直接投資（FDI）、兼併（M＆A）、跨國公司、比較利益、貿易與投資的替代效應。

台灣電線電纜產業在面臨市場仍然處於供過於求、粥少僧多的局面；加上台灣已加入WTO世貿組織，以及未來可能與中國簽署ECFA協定，即將面對開放國內市場的衝擊下；對電線電纜企業而言，實刻不容緩必須盡速建立非價格條件的核心競爭力。本研究期以供應鏈管理理論做基礎，並參考台灣其他傳統產業已導入供應鏈模式的實務經驗，以台灣電力線纜民間市場為範圍，藉由針對現行業者A公司的經營實務進行研究分析，希望能找出適用於此產業的供應鏈模式。 本研究從競爭策略理論切入，探討企業面臨生存競爭的抉擇，以五力分析開始，再探討供應鏈管理觀點，引導企業經營模式變革的方向，加上對成功案例的分析，以研究產業基本供應鏈運作模式。 其次，基於能夠見樹又見林的概念下，從台灣電線電纜產業整體概況出發，進而深入解讀台灣電力線纜產業的特性及概況，最後鎖定在台灣電力線纜民間市場的實務特性，以期發現當前產業及企業所面臨的潛在經營難題，作為本研究內在及外在環境分析的基礎。然後，藉由對A公司個案的研究，以供應鏈管理理論基礎及其他產業成功案例的實務經驗，針對台灣電力線纜產業及A公司的經營現況進行分析，確立對於建立供應鏈模式的必要動機；並找出必要的考量因素；進一步對實務面進行解析，以發現潛在問題；並針對問題找出可行方案及評估其效益。 最後，針對本研究做出綜合性的結論，以及對台灣電力線纜產業與個案公司提出管理建議；並對於後續有興趣研究相關議題的研究者，提供研究方向的建議及參考。 本研究並不冀望能為台灣電力線纜產業提出一套完整的供應鏈模式；但期望能藉本研究作開路先鋒，先求有，再求好，建立基礎模式；並希望引導後續更多相關產官學先進的投入研究改善，讓電線電纜產業注入創新概念活水，以培養新的經營概念，提升產業整體實力，以求達到永續經營的目標。

本研究目的旨在探討國中教師甄選口試，背景變項（口試委員特徵、應試者特徵、應試者學經歷）及口試委員心理因素（應試者儀容舉止、應試者受喜愛程、應試者教師專業、口試委員類我效應），如何影響國中教師甄選口試評價結果以及這些因素與口試評價結果間的因果關係模式。 本研究採用問卷調查法進行，研究工具為「口試評量表」。資料蒐集的對象以93年度台北縣、台北市及桃園縣擔任國中教師甄選之口試委員為對象。參與本研究之口試委員共78位，每位口試委員口試應試者的人數3至23人不等，共回收口試評量表883份，有效量表844份。每位口試委員填答之量表以10份作為本研究原始資料選取之基準，共選入有效量表570份進行資料分析。 根據資料分析的結果，本研究獲致結論如下：（1）口試委員口試經驗、教育程度、職務，應試者教育背景、任教領域等背景變項對口試評價結果有顯著影響。（2）本研究建構之「國中教師甄選口試決策模式」，獲得實證資料驗證。模式中各心理因素對口試評價結果之總效果以口試委員類我效應最強，其餘依次是應試者受喜愛程度、應試者教師專業、應試者儀容舉止。 / The purpose of my study aimed at investigating how the background variables (interviewer characteristics, applicant characteristics, the educational background and past experience of the applicants), the interviewer’s psychological factors (the appearance and behavior of the applicants, the likability of applicants, teacher specialty of the applicants, and the perceived similarity of the interviewer), influence the interview assessment of the teacher selection in junior high school. And how the relations of cause-effect modeling between those mentioned-above factors and the result of interview assessment are formed. The survey is used and the instrument is the Interview Assessment Form. The objects of data-gathering are those who are the interviewers of junior high school teacher selection in Taipei County, Taipei City, or in Taoyuan County, in 2004. A total of 78 interviewers participate in my research, and each interviewer has 3 to 23 interviewees. Thus, a total of 883 Interview Assessment Forms are collected, 844 of which are effective forms. The average ten Assessment Forms answered by each interviewer are chosen as the standard of the choice of original data, and a total of 570 effective Assessment Forms are selected for data analysis. According to the result of data analysis, the conclusions of my research are as follows: (1) Such background variables as the interviewer’s interview experience, level of education, job position, the educational background of the applicants, and teaching fields have significant influences upon the result of interview assessment. (2) The decision model of the interview of junior high school teacher selection well developed in study is verified by well-examined data. From the analysis of Structural Equation Modeling, the strongest influence of the psychological factors upon the total effect of interview assessment is the interviewer’s perceived similarity, followed by the likability of applicants, teacher specialty of the applicants, and the appearance and behavior of the applicants.

推薦系統的出現是為了解決訊息過載的問題，其需求隨著科技的進步、網路的普及而增加，相關技術也越發多樣且成熟。廣泛應用於各領域的統計模型也在技術的行列中。 推薦系統的運作仰賴使用者偏好訊息，而使用者對項目所組成的偏好空間往往十分巨大且不平衡，統計上需要相對複雜的隨機效應模型或混合效應模型來描繪這樣的變數結構，且通常需要計算效率相對低的反覆疊代過程來估計模型參數。因此Perry（2014）、Gao & Owen（2016）先後提出以動差法處理階層線性模型與兩因子隨機效應模型，是一種犧牲統計效率換取計算效率的做法。 本研究便是採用統計模型中的隨機效應模型法，分別以最大概似法和動差法估計參數，與同為協同過濾技術觀點的矩陣分解法進行分析比較。透過預測準確度和運算效率兩個層面，來評估各演算法在MoiveLens這筆資料上的推薦表現。 根據試驗結果歸納出隨機效應模型法無論以什麼樣的參數估計方式，在預測準確度的表現上都不如矩陣分解法來得好；但以動差法估計參數在穩定度上與矩陣分解法的表現差不多，且在運算效率上好很多。 / The recommender system (RS) appeared to solve the problem of information overload. The demand of the RS has increased with the advancement of technology and the popularity of the Internet, and related techniques have become more diverse and mature. The statistical models widely used in various fields are also in the list of techniques. The operation of the RS relies on user preference information, and the space of users’ preference to items is often large and unbalanced. Statistically, relatively complex random effects models or mixed effects models are needed to describe such variable structures, and often require a large number of iterations to estimate model parameters. Perry (2014), Gao & Owen (2016) proposed using the moment-based method to deal with hierarchical linear models and two-factor random effects models, respectively, expressing an idea of sacrificing statistical efficiency in exchange for computational efficiency. In this study, we analyze and compare the random effects model, using the maximum likelihood method and the moment-based method to estimate the parameters with the matrix factorization. Through the prediction accuracy and computational efficiency to evaluate the performance of each algorithm on the MoiveLens data. According to the experiment results, the random effects model is not as good as the matrix factorization in terms of the prediction accuracy no matter what kind of parameter estimation method is used; however, the performance of the moment-based parameter estimation is consistent with the matrix factorization in terms of the prediction stability, and much better in terms of the efficiency.

疤痕效應(Scarring Effect)所指的當一個人的就業歷程出現失業，失業將會對往後的就業歷程產生長期性的影響，且影響的範圍相當廣，包含收入、就業狀況、身心健康等等。然而，在所有年齡層中，青年是最容易失業且待在勞動市場中時間最長的群體。 　　本研究將在疤痕效應的理論基礎上探討：台灣青年在失業後是否會影響終生的就業歷程、造成長期性工作貧窮的現象。本研究試著從青年找不到適得其所的工作時，其實際工作狀況和生活來理解失業的成因及長期性影響。 　　本研究採取多元方法的研究設計，透過「人力運用擬-追蹤調查資料庫」以及「華人家庭動態資料庫」比較15-29歲有失業經驗的青年及無失業經驗青年，往後薪資及就業狀況，觀察失業的短期性及長期性的影響。並以深度訪談的方式，訪談七位不同背景且在15-29歲期間有3個月以上失業經驗的受訪者，透過訪談青年失業理解連續性脈絡，藉此理解疤痕效應的成因及疤痕效應造成影響的詳細情形。 　　實證結果發現不論觀察期長短，有失業經驗的青年的收入漲幅普遍會低於沒有失業經驗的青年，失業確實會對收入水準造成長期影響。就業情形的差異上，有失業經驗青年陷入再度失業的機率普偏高於無失業經驗青年陷入失業的機率。 　　訪談結果發現疤痕效應的原因有(一)工作經驗不足，無法進入職場；(二)雇主對過短工作經驗的負面評價；(三)人力資本難以積累；(四)社會關係封閉，難以向外得到援助；(五)工作經驗的中斷。 　　本研究透過疤痕效應的理論，指出失業對青年所造成的長期性影響。藉由實證數據及訪談結果，深入理解青年失業所造成的長期性影響及成因，作為解決青年失業問題及工作貧窮上的政策參考。 / The “Scarring Effect” means the unemployment of a person's career path, which will have a long-term impact on the future career path, and have the influence of a wide range, including income, employment status, physical and mental health and so on. However, youth have the highest unemployment rate in all ages and have the longest time in the labor market. 　　This study will explore the theory of scarring effect: whether unemployment will affect Taiwanese youth’s lifetime career path after unemployment, resulting in long-term poverty. This study tries to understand the causes and long-term effects of unemployment from the time when youth find jobs that are not suitable for them. 　　This study is based on a multimethod research approach to compare youth from 15 to 29 years old with unemployed experience and young people with no unemployment experience. Through "Manpower Utilization Quasi-longitudinal Survey" and the “Panel Study of Family Dynamics" in subsequent earnings and employment status, we observe the short-term and long-term effects of unemployment. And through deep-interviews with seven different backgrounds and those who had more than three months of unemployment experience during the 15-29 year old period, we understand the continuity of unemployment, the causes of scar effects and the details of the impact. 　　The empirical results show that regardless of the length of the observation period, the income growth of youth with unemployed experience is generally lower than which of those without unemployed experience. Unemployment will indeed have a long-term impact on income levels. The difference in employment status, the re-unemployment rate of the youth with unemployment experience is higher than which of youth without unemployment experience. 　　The results of interviews finds that the cause of scarring effects are contributed to (1)Lack of working experiences to get work. (2)Employer have negative impression on short working experience.(3)The difficulty of accumulating human capital.(4) Closed social relationships, which is difficult to get assistance from others. (5)Interruption of work experience. 　　This study, through the theory of scarring, points out the long-term effects of unemployment on youth. Through empirical data and interviews, we will understand the long-term impact and causes of youth unemployment as a policy reference to solve the problem of youth unemployment and working poverty.