As for the calculation of user simi

As for the calculation of user similarity, the thesis primarily defines a two-dimensional matrix top20booknums[users.Count,20] with users.Count rows and 20 columns to store 20 users with maximum similarity. The similarity is progressively sorted from high to low. Meanwhile, a two-dimensional matrix top20cardnums[users.Count,20] is defined to store user information in the two-dimensional matrix top20booknums. Next task is to successively find most similar 20 neighbors for each user. User similarity is measured by the number of similar borrowed books. Since above processes have filled in reader borrowing records in the two-dimensional borrowing matrix wherein 1 means ‘‘having borrowing records’ and o means ‘‘no borrowing records’’, this section will calculate user I and user ’s similarity with formula (2). As shown in the following Fig. 4, if the thesis plans to calculate the similarity between reader 1 and 2, then we could regard reader 1’s grade vector as u1 = (1,0,0,1,1) and reader 2’s grade vector as u2 = (1,1,1,1,1). The similarity between reader 1 and 2 could be calculated pursuant to above formula. By that analogy, the thesis could finally obtain the most similar 20 neighbors of every user.