Similarity analyses are most likely to involve the greatest overhead, especially in social media where datasets tend to be large; scalability is a computational expense for large datasets so require more efficient approaches. For such cases, techniques (such as the “expectancy violations theory,” which looks for deviations from a baseline) may be an efficient way to filter suspect cases for further examination. This is a computationally cheaper alternative that can be applied to both sender and content deception; for example, comparing deviations from a normal user baseline requires parsing a database just once, leading to a complexity of O(N).