全球智慧型手機的出貨量持續提升，且熱門市集的App下載次數紛紛突破500億次。而在iOS和Android手機App市集中，App的評價和評論對App在市集的排序有很大的影響；對於App開發者而言，透過評論確實可掌握使用者的需求，並在產生抱怨前能快速反應避免危機。然而，每日多達上百篇的評論，透過人力逐篇查看，不止耗費時間，更無法整合性的瞭解使用者的需求與問題。 文字情感分析通常會使用監督式或非監督式的方法分析文字評論，其中監督式方法被證實透過簡單的文件量化方法就可達到很高的正確率。但監督式方法有無法預期未知趨勢的限制，且需要進行耗費人力的文章類別標注工作。 本研究透過情感傾向和熱門關注議題兩個面向來分析App評論，提出一個混合非監督式與監督式的中文情感分析方法。我們先透過非監督式方法標注評論類別，並作視覺化整理呈現，最後再用監督式方法建立分類模型，並驗證其效果。 在實驗結果中，利用中文詞彙網路所建立的情感詞集，確實可用來判斷評論的正反情緒，唯判斷負面評論效果不佳需作改善。在議題擷取方面，嘗試使用兩種不同分群方法，其中使用NPMI衡量字詞間關係強度，再配合社群網路分析的Concor方法結果有不錯的成效。最後在使用監督式學習的分類結果中，情感傾向的分類正確率達到87%，關注議題的分類正確率達到96%，皆有不錯表現。 本研究利用中文詞彙網路與社會網路分析，來發展一個非監督式的中文類別判斷方法，並建立一個中文情感分析的範例。另外透過建立全面性的視覺化報告來瞭解使用者的正反回饋意見，並可透過分類模型來掌握新評論的內容，以提供App開發者在市場上之競爭智慧。 / While the number of smartphone shipment is continuesly growing, the number of App downloads from the popular app markets has been already over 50 billion. By Apple App Store and Google Play, ratings and reviews play a more important role in influencing app difusion. While app developers can realize users’ needs by app reviews, more than thousands of reviews produced by user everday become difficult to be read and collated. Sentiment Analysis researchs encompass supervised and unsupervised methods for analyzing review text. The supervised learning is proven as a useful method and can reach high accuracy, but there are limits where future trend can not be recognized and the labels of individual classes must be made manually. We concentrate on two issues, viz Sentiment Orientation and Popular Topic, to propose a Chinese Sentiment Analysis method which combines supervised and unsupervised learning. At First, we use unsupervised learning to label every review articles and produce visualized reports. Secondly, we employee supervised learning to build classification model and verify the result. In the experiment, the Chinese WordNet is used to build sentiment lexicon to determin review’s sentiment orientation, but the result shows it is weak to find out negative review opinions. In the Topic Extraction phase, we apply two clustering methods to extract Popular Topic classes and its result is excellent by using of NPMI Model with Social Network Analysis Method i.e. Concor. In the supervised learning phase, the accuracy of Sentiment Orientation class is 87% and the accuracy of Popular Topic class is 96%. In this research, we conduct an exemplification of the unsupervised method by means of Chinese WorkNet and Social Network Analysis to determin the review classes. Also, we build a comprehensive visualized report to realize users’ feedbacks and utilize classification to explore new comments. Last but not least, with Chinese Sentiment Analysis of this research, and the competitive intelligence in App market can be provided to the App develops.

如何從大量的資料中擷取隱匿或不容易直接觀察的資訊，是重要的議題，社會網路提供了一個適當的系統描述模型與內部檢視分析的方法，過去社會網路分析多著重於靜態的分析，無法解釋發生在網路上的動態行為；我們的研究目的是從動態社會網路分析的角度，觀察政治權力的變化，將資料依時間切分成多個資料集，在各個資料集中，利用官員共同異動職務及共事資料建構網路，並使用EdgeBetweenness分群方法將網路做分群，以找出潛在的政治群組，接著再採用事件導向的方法(Event-based Framework)，比較連續兩個時間區間的網路分群結果，以觀察政治群體的動態發展，找出政治群組事件，並將其匯集成政治群組指標，以用來衡量政治群組的變動性及穩定性。我們提供了一個觀察政治權力變化的模型，透過網路建立、網路分群到觀察網路動態行為，找到不容易直接取得的資訊，我們也以此觀察模型解決以下問題：(1)觀察部門之接班梯隊之變化，(2)觀察特定核心人物之核心成員組成模式，(3)部門專業才能單一性或多元性之觀察。實驗結果顯示，利用政治群組事件設計的政治群組指標，可實際反應政府部門選用人才的傾向為內部調任或外部選用。 / Extracting implicit information from a considerable amount of data is an important intelligent data processing task. Social network analysis is appropriate for this purpose due to its emphasis on the relationship between nodes and the structure of networked interactions. Most research in the past has focused on a static point of view. It can't account for whatever action is taking place in the network. Our research objective is to observe the evolution of political power by dynamic social network analysis. We begin by creating static graphs at different time according to the synchronous job change between the government officials or the relationship between the government officials whom work in the same government agency. We obtain political communities from each of these snapshot graphs using edge betweenness clustering method. Next we define a set of evolutionary events of political communities using event-based framework. We compare two consecutive snapshots to capture the evolutionary events of political communities. We also develop two evolutionary political community metrics to measure the stability of political communities. We propose a model of observing the evolution of political power by three steps－network construction, community identification and community evolution tracking. The approach is shown to be effectual for the purposes of: (1) finding succession pool members in government agencies, (2) observing the inner circle of a leading political figure, (3) measuring the specialized degree of government agencies. Experiments also show that our community evolution metrics reflect the tendency of whether a government agency conducts internal succession or outside appointment.

對於零碎且結構複雜的資料來源時，社會網路分析能夠給予整體性的觀察，還能檢視個體之間的關係。目前社會網路分析研究中，因為將網路退化至簡單連結關係，所以會遺失許多珍貴的資訊。而網路規模和型態隨著研究議題的不同，也會跟著增大或趨於多變，但動態網路分析能夠提供我們檢視每個時期，網路的變化或社群的形成或消失，甚至能知道節點間的互動影響。本論文研究，以政府人事異動資料為主，並且加入了其他政府組織的相關資料，建構出政府組織的從屬網路，並在每個網路快照中，擷取出重要的官員異動；每一筆人事異動都是一個事件的發生，而特任或簡任官員在本研究中視為重要事件，從這些重要事件的發生，我們能夠對每個時間的官員，使用EventRank的演算法做排名計算。最後能從時間的變化中，觀察出每個時期的佔有重要影響力的官員。 / To fragmented and complex structure data, social network analysis (SNA) can give an overall observation, but also view the relationship between individuals. Recent research in SNA is the degradation of the network link to a simple relationship but it will lose a lot of valuable information. The size and type of network with different research topics will follow the increase or rapidly changing, dynamic network analysis can provide our view of changes in the network or community to form or disappear in every period, even know the impact of the interaction between nodes. This thesis is based on the government official changes and other related data to construct manager-subordinate network of the government organization and capture the important interactions between officials in every network snapshot. An official change is the occurrence of an event and special level official changes in this study as a critical event. From these critical events, we can use the Event Rank algorithm to rank the officials. Finally, we can observe which official has more influence from the time changes.