Research Goal Develop an influence diffusion detection model to automatically detect the influence diffusion between two linked bloggers. Research Phase 1 Research Method: Apply statistical methods to determine blog features suitable for detecting influence diffusion. Output: Similar sentiment polarity expressed on common topics significantly indicate influence diffusion between the linked blog posts. Research Phase 2 Research Method: Apply sentiment analysis approach based on CSR automatically generated typed dependencies rules and a subjective phrase lexicon approach. Output: A Sentiment Analysis model based on typed dependencies rules and subjective phrase lexicon, which takes into account the complex relationships between words. Research Phase 3 Research Method: Develop influence diffusion detection model based on the Independent Cascade and Linear Threshold approach using sentiment values between the linking and linked bloggers, agreement towards the linked bloggers, and influence styles as features. The Sentiment Analysis model from phase 2 was applied to detect the sentiment and agreement values between the linking and linked bloggers. Output: The Influence Diffusion Detection model to detect influence diffusion between the linking and linked bloggers. Research Objective (RO1) Identify the relevant blog features that could be used to detect influence diffusion between linked blog posts. Research Method (RM1) Perform statistical analysis to determine blog features suitable for detecting influence diffusion. Research Question (RQ1) What are the relevant blog features that could be used to detect influence diffusion between linked blog posts? Research Objective (RO2) Establish a sentiment detection approach that takes into consideration the complex relationships between words in the blog posts to automatically detect influence diffusion between linking and linked blog posts. Research Gap (RG2) Existing heuristic approach to sentiment analysis is based on available knowledge base where rules generation is not automatic, and may not provide a complete set of rules. Research Question (RQ2) How can the sentiment analysis process be enhanced to better detect the sentiments expressed within blog posts through considering the complex relationships between words? Research Method (RM2) Apply sentiment analysis based on CSR automatically generated typed dependencies rules and a subjective phrase lexicon approach. Research Objective (RO3) Develop an influence diffusion detection model using sentiments on common topics between linking and linked bloggers, agreement towards the linked bloggers, and bloggers' influence styles as possible features to improve influence diffusion detection between linked bloggers. Research Gap (RG3) Influence diffusion could not be described using a single detached feature based on the existence of links or similar sentiment polarity. Research Question (RQ3) Will automatic influence style analysis help detect influence diffusion between linked bloggers? Research Method (RM3) (i) Determine eng...
