ERP leverantörer arbetar ständigt med innovation för att vara konkurrenskraftiga. För att ständigt hålla sig konkurrenskraftiga måste företag arbeta med utveckling av ERP system, Machine Learning, big data och analytics. Att använda dessa tekniker i en kombination hjälper företag att kunna utveckla automatiserade funktioner för kunder. Denna studie är genomförd utifrån en kvalitativ ansats där vi analyserar två olika företag som levererar ERP system i olika branscher. Fokuset kommer att ligga på fenomenen maskininlärning, analytics och ERP system. I denna studie skapas en förståelse för viktiga begrepp och hur de fungerar tillsammans. Men även hur ERP leverantörer arbetar med ERP system, maskininlärning och analytics för att sedan kunna se om det är möjligt att anpassa systemets passform med hjälp av maskininlärning. Studien visar att det finns hinder som företag måste hantera när de arbetar datadrivet. / ERP suppliers are constantly working to be more innovative and to be more competitive. To be able to constantly stay competitive, companies must keep on working with development of ERP systems, Machine Learning, Big Data and Analytics. To use these techniques in combination with each other helps organisations to keep on developing automated functions for their customers. This study was conducted on the basis of a qualitative approach where we analyze two different organisations that deliver ERP systems to different industries. The focus in this study is applied to the concepts of Machine Learning, Analytics and ERP systems. This study also creates an understanding of these important concepts and how they work together. But also how ERP suppliers work with ERP systems, Machine Learning and analytics to be able to see if it is possible to create a better fit for the system with Machine Learning. This study also shows that there are obstacles that organisations must deal with when they work data-driven. This study is written in Swedish.

The thesis is focused on the communication inclusion and exclusivity of Czech journalist on Twitter and how they use conventions of this platform to connect with other users. Through the description of current communication layers and functions of Twitter this thesis depicts how it became one of main sources of news and how it pushed journalists to reinterpret their traditional roles in the society. It also describe how digital humanities and digital trace data gathered from social media can be used as means of analysis of social interactions of its users. The practical part presents a cluster analysis based on Twitter data of 457 Czech journalists that shows how is Twitter used to communicate within and outside the Czech media system.

現今因無線通訊技術的進步，使得人們能方便地利用智慧型裝置透過3G，4G和Wi-Fi等技術彼此溝通聊天。其中，聊天應用是最受智慧型裝置使用者歡迎的應用程式。大部分的聊天應用程式需依賴網路以達到訊息交換的目的。然而網路的頻寬是非常有限的，當使用者處在擁擠的環境中時，他們可能會面臨資源耗盡問題。此外，例如在漫遊的情況下有些使用者並沒有行動網路的存取，導致使用者無法使用聊天應用。 因此我們希望利用無線廣播傳輸的特性，開發一個應用於間歇性網路連接的聊天應用程式。然而，廣播傳輸的散播策略若沒有設計得宜，可能導致廣播風暴的問題，使得整體網路效能低落。我們研究的目標是要如何在間歇性網路增加訊息的傳輸效率。為了達成此目標，在我們的研究中考量了許多技術要求，如：訊息具有截止時間與優先權特性、多聊天室應用、傳輸效率。 我們提出了一種兩階段基於網絡編碼設計的訊息散播方法，實現在機會性社群網路中的訊息散播。網絡編碼階段，提高網路頻寬的傳輸效率，也能增加網路傳輸的可靠性；預熱階段能提升網路編碼訊息被解開的機率。最後，利用政大的真實軌跡紀錄評估我們所設計的訊息傳播方法。結果顯示，我們的方法是有效率且優於氾濫式的路由協議和一般的網絡編碼散播技術。 / Nowadays, the advancement of wireless communication technology has allowed people to use smart phones to communicate with each other more easily via 3G/4G, Wi-Fi, etc. One kind of popular mobile Apps is “chat” App. Most chat Apps rely on the Internet to exchange the messages. However, the bandwidth of network is limited in some circumstances. When users stay in the crowded environment, they will face the resource depletion problem. Besides, some people may not subscribe to any cellular network access, e.g. in roaming scenarios. Therefore, we want to develop a novel mobile Chat APP in intermittently connected networks. We utilize the characteristic of the wireless broadcast transmission. However, it may cause the broadcast storm problem without careful design. How to increase the efficiency of message delivery in such intermittently connected networks is our research goal. To achieve this, technical issues in our research involve message priority, multi-chatroom, deadline and transmission efficiency. We proposed a two-phase network coding design for message dissemination to enable the multi-hop instant messaging in Opportunistic Mobile Social Networks. The network coding phase can increase the bandwidth utility and transmission efficiency. Moreover, it can improve transmission robustness and adaptability. The warm up phase can increase the decoding probability of coded packets. Finally, we evaluated our approach with real trace data from NCCU. The results showed that our approach is effective and superior to the flooding based routing protocol and the pure network coding technique.