Content-based recommender systems recommend unseen items to the active user based on the user's preferences of the past. The simplest way to find which items are the most similar to the one that the user has liked in the past is to compute the similarity of an unseen item with the user profile based on the keyword overlap. User profile Title Genre Fiction Author Ahmad Patria Abdullah Type Paperback Keywords detective, murder, New York Item representation Title Genre The Night of Memoir the Gun Author David Carr Type Paperback The Lace Reader Fiction, Mystery Brunonia Barry Hardcover Keywords Press and journalism, drug addiction, personal memoirs, New York American contemporary fiction, detective, historical American fiction, murder, neo-Nazism Into the Fire Hardcover Romance, Suspense Suzanne Brockmann Figure 1 Based on Figure 1, answer the following question. a. Using the following Dice coefficient, find the similarity of all the unseen items with user's profile. [keywords(bi) keywords(b;) Dice coefficient = 2 x keywords(bi)+keywords(b;)|| where, keywords (b;) are the keywords for book b; and keywords(bi) are the keywords in the user profile. (8 Marks) b. Keywords usually don't give good results. How can we improve the results based on the user profile and item representations? (2 Marks)