The heuristic rules we used in our implemented algorithm may be summarized as follows: a component with a meaningful ID spelling will be assigned a relative higher score, a component with too many numbers (>=4) will be assigned a relative lower score, a component with an ID containing “middle” has relative higher score than “top” and “bottom”, a component with an ID containing “footer”, “left”, and “right” will be assigned a relative lower score, and a component with an ID containing advertisementoriented words will be assigned a relative lower score
This score algorithm will provide clients with a score list associated with the component IDs. Based on these scores, clients can make their decision to keep or eliminate in a more precise manner. They can even choose to use a score filter to auto-select components, e.g. if the client set a 30 score filter for a certain URL, all components with scores lower than 30 will be automatically deleted when the mobile-optimized web page is generated.