The recommendation processor may then select a weight for the pairing of the selected content and bias that weighting based on the calculated time difference (block 516). As described above, biasing the weight may include increasing the weight as the time difference approaches a multiple of a first predetermined time period (e.g. 24 hours) and decreasing the weight as the time difference departs from a multiple of the first predetermined time period (e.g., 24 hours). Biasing the weight may also include increasing the weight further, i.e., to a greater extent when the time difference approaches a multiple of a second predetermined time period that is also a multiple of the first predetermined time period. As noted above, the first predetermined time period may be 24 hours, and the second predetermined time period may be 168 hours (e.g., a week long time difference). Stated differently, a time-series of weight values plotted with respect to time difference may exhibit increasing weight values as the time difference approaches a multiple of the first predetermined time period (e.g., approaches 24 hours) and exhibit a peak weight value where the time difference is exactly a multiple of the first predetermined time period (e.g., is exactly a multiple 24 hours). The time-series of weight values may also exhibit decreasing weight values as the time difference departs from a multiple of the first predetermined time period (e.g., departs from a multiple of 24 hours) and exhibit a valley weight value where the time difference is exactly between two multiples of the first predetermined time period (e.g., exactly between two multiples of 24 hours, i.e., a multiple of 24 hours plus 12 hours). The weight values may then begin to increase again as the time difference departs from this middle point and again approaches the next multiple of the first predetermined time period (e.g., 24 hours). Since the weight value correlates with the time difference, in this example, the proposed approach provides improved recommendations even when the time difference isn't exactly a multiple of the first predetermined time period (e.g., 24 hours), for example, where the time difference is a multiple of the first predetermined time period plus or minus x number of minutes (e.g., ?0.5 hours, +0.5 hours, etc.). Biasing the weight may further include applying a decay factor to the weight. As described above, the decay factor, in some example implementations, may be inversely proportional to the time difference such that the decay factor decreases as the time difference increases.