Loading...
Modeling and Analysis of Information Diffusion over Social Networks using Point Processes
Jafarzadeh, Sina | 2016
825
Viewed
- Type of Document: M.Sc. Thesis
- Language: Farsi
- Document No: 48832 (19)
- University: Sharif University of Technology
- Department: Computer Engineering
- Advisor(s): Rabiee, Hamid Reza
- Abstract:
- Social networks play an important role in the spread of information, news and users’decisions. An important type of these networks are location-based networks where users can post the time and location of their checks in different places. Check-ins of a user may incentivize her friends to visit the same location in the near future.This phenomena is called diffusion process. Previous works often assume that the generation of an event just increases the probability of subsequent events in a near future. Even though the assumption is quite valid in the social networks like Twitter,it is not the sufcient assumption in spatio-temporal social networks where the users’physical limitations and patterns are considerable too. For instance, people go (and check-in) in restaurants every 24h to have their dinner meals. Therefore, generation of an event in time t should increase the probability of another event being generated in time t+24. Besides, the previous works consider separate cascades for different contagions in the network. Taking a different cascade for each unique location in spatio-temporal social networks with ∼10000 unique locations is not possible. The increased number of parameters make the model computationally intractable and inefcient. We propose a two-level hierarchical model for diffusion processes on location-based social networks based on stochastic point processes. Our generative model has two parts: temporal and spatial. In the temporal part, we generate the time and category of check-ins using a periodic point process which models the periodic pattern in the behavior of users. In spatial part, given the time and category of all check-ins, we choose the location of them with a probability which is higher for the recent locations visited by the user or her friends. By decomposition of the problem to spatial and temporal levels, the number of parameters is reduced considerably. In order to learn the model parameters, we solve a maximum likelihood problem using the expectation maximization and convex optimization algorithms. Experiments on synthetic and real data show that the proposed method outperforms the baselines in terms of the likelihood and prediction quality of time and location. The results also verify the previous assumption that the users’ check-in times follow a periodic pattern and the location of her check-ins depends on the past activities of user and her friends
- Keywords:
- Complex Network ; Information Diffusion ; Point Process ; Spatiotemporal Patterns ; Social Networks ; Stochastic Process ; Location-Based Networks
-
محتواي کتاب
- view
- فهرست شکلها
- فهرست جدولها
- مقدمه
- فرآیند انتشار
- فرضیات در مدلهای انتشار
- تعریف مسأله
- چالشها و زمینههای تحقیق
- مدلسازی مستقل از اطلاعات جانبی
- مدلسازی وابسته به اطلاعات جانبی
- راهکار پیشنهادی و دستاوردهای رساله
- جمعبندی
- فرآیند انتشار
- روشهای پیشین
- مدلسازی مستقل از اطلاعات جانبی
- روشهای کلاسیک
- روشهای مبتنی بر فرآیندهای تصادفی
- مدل وابسته به اطلاعات جانبی
- جمع بندی
- مدلسازی مستقل از اطلاعات جانبی
- روش پیشنهادی
- مبنای نظری
- فرآیندهای تصادفی دورهای
- مدل پیشنهادی
- استنتاج
- جمعبندی
- نتایج آزمایشها
- مجموعه داده
- مجموعه داده ساختگی
- مجموعه داده واقعی
- روشهای مورد مقایسه
- معیارهای سنجش
- جزئیات پیادهسازی
- نتایج ارزیابی
- دادههای ساختگی
- دادههای واقعی
- جمعبندی
- مجموعه داده
- جمعبندی و کارهای آتی
- پیوست الف ; فرآیندهای تصادفی نقطهای
- تابع شدت شرطی
- پیوست ب ; مشتقات در بهینهسازی مدل مکانی-زمانی
- پارامترهای مربوط به زمان رویدادها
- پارامترهای مربوط به مکان رویدادها
- مراجع
- واژهنامه انگلیسی به فارسی
- واژهنامه فارسی به انگلیسی