Citation Context

...nsider a user as a query term. Therefore, wesemploy the Normalized Discounted Cumulative Gain (NDCG)s1 In this paper, we do not present the questions of the onlinessurvey due to the space limitation.s=-=[11]-=- to measure the list of recommended potential friends. Forseach list of recommended potential friends, we can obtain a scoreslist where the scores are provided by ground truth. Such a list isscalled r...

Citation Context

...le, a user regularly goes to thesschool, but sometimes passes by a gas station. Hence, to identifysthe user frequent movement behaviors, we perform the sequentialspattern mining algorithm Prefix-Span =-=[10]-=- on each user’sssemantic trajectory dataset to mine the frequent semanticstrajectories. Take Figure 6 as an example. Given the Trajectory1sand Trajectory2 are from a mobile user, her trajectory log wi...

Citation Context

...t-similar users to user1.s4. EXPERIMENTSsIn this section, we conduct a series of experiments to evaluatesthe performance for the proposed friend recommendation systemsusing MIT reality mining dataset =-=[5]-=-. All the experiments aresimplemented in Java JDK 1.6 on an Intel Core Quad CPU Q6600s2.40GHz machine with 1GB of memory running MicrosoftsWindows XP. We first present the data preparation on the MITs...

Citation Context

... in Section 4.sThe conclusions and future work are given in Section 5.s2. RELATED WORKsMany studies had discussed the similarity measurementsproblems in data mining. Trajectory similarity measurement =-=[6]-=-sand user similarity measurement [7][8][12] are two hot topics insthis problem domain. In [6], Lee et al propose a Partition-andGroup method to calculate the similarity between two trajectories.sThey ...

Citation Context

...sue of measuringsmobile users’ similarity in terms of their trajectories has beensdiscussed in the literature, existing studies mostly focus only onsanalyzing geographic features of user trajectories =-=[7]-=-[8][12]. Assmentioned earlier, a geographic trajectory typically consists of assequence of geographic points (represented as <latitude,slongitude>), tagged with timestamps. As a result, thesmeasuremen...

Citation Context

...rajectories, the physical behaviors of users can besextracted from user trajectories.sWith the development of Web 2.0 technology, many mobilesusers are willing to share their trajectories with others =-=[12]-=-. Asnumber of forums have been established to facilitate sharing ofstrajectories among their users [1][2]. We envisage that such logssof user trajectories will also be available and sharable in manysL...

Citation Context

...hat merely using the geographic information to capture thestrajectory similarity as well as user similarity is not sufficient.sThe notion of semantic trajectory has been proposed bysAlvares et al. [3]=-=[4]-=-. Basically, a semantic trajectory consists of assequence of locations with semantic tags to capture the landmarksspassed by. Consider Figure 2 where trajectories are tagged with asnumber of semantic ...

Citation Context

...k},s{Hospital}> and all of its subsequences will be mined.sHowever, it is clear to observe that the longer pattern we minesthe more subsequences will be generated due to the downwardsclosure property =-=[9]-=-[10]. It leads to biased measure of users’ssimilarity, because all the subsequences of a pattern will besinvolved in the user similarity calculation. For example, thessubsequences of the pattern <{Unk...

Citation Context

... of measuringsmobile users’ similarity in terms of their trajectories has beensdiscussed in the literature, existing studies mostly focus only onsanalyzing geographic features of user trajectories [7]=-=[8]-=-[12]. Assmentioned earlier, a geographic trajectory typically consists of assequence of geographic points (represented as <latitude,slongitude>), tagged with timestamps. As a result, thesmeasurement o...

Citation Context

...e that merely using the geographic information to capture thestrajectory similarity as well as user similarity is not sufficient.sThe notion of semantic trajectory has been proposed bysAlvares et al. =-=[3]-=-[4]. Basically, a semantic trajectory consists of assequence of locations with semantic tags to capture the landmarksspassed by. Consider Figure 2 where trajectories are tagged with asnumber of semant...