Software developers have benefited from various sources of knowledge such as forums, question-and-answer sites, and social media platforms to help them in various tasks. However, extracting software-related knowledge from different platforms involves many challenges. Our work is motivated by the following use cases: (1) We consider a developer who wants to acquire new knowledge based on software-relevant tweets. Developers face challenges while using Twitter, which relate to having to deal with a huge amount of irrelevant tweets produced on Twitter. As the developer has limited time to inspect new tweets, a content aggregator for Twitter data related to software development is essential. More generally, automatic identification of software-relevant tweets will also enable downstream applications such as creation of a specialized Twitter feed for the developer community. (2) We consider a software developer who acts as a content creator and publishes a coding tutorial on YouTube. Viewers can visually follow the instructions provided in the videos and leave a comment that expresses their experience with the video. Digesting information taken from the comments will help the content creator to be more engaged with their audience and improve their future videos. Automatic filtering of relevant comments will enable creators to study feedback provided by viewers more efficiently, and similar to tweets, such filtering can be used to improve downstream analytics tasks, such as detection of common topics among relevant comments. In both platforms mentioned above (Twitter and YouTube comments), sentences are typically short, contain a lot of noise, and may contain non-standard words. In order to address these challenges, we propose SIEVE~\cite{sulistya2019sieve}, an approach to improve the effectiveness of knowledge extraction tasks by performing cross-platform analysis. Our approach is based on transfer representation learning and word embedding, leveraging information extracted from a source platform which contains rich domain-related content. The information extracted is then used to solve tasks in another platform (considered as target platform) with less domain-related content. We first build a word embedding model as a representation learned from the source platform, and use the model to improve the performance of knowledge extraction tasks in the target platform. We experiment with Software Engineering Stack Exchange and Stack Overflow as source platforms, and two different target platforms, i.e., Twitter and YouTube. We conducted experiments based on the existing datasets provided by Sharma et al. for Twitter, and Poche et al. for YouTube comments. Our experiments show the effectiveness of our proposed cross-platform analysis approach which achieves performance improvements of up to 28% and 10.3% for the first and second use case respectively.