Surety is performing system maintenance this weekend. Electronic date stamps on new Prior Art Database disclosures may be delayed.
Browse Prior Art Database

System and architecture for Mining SMS Messages on Mobile Devices Preserving Users' Privacy

IP.com Disclosure Number: IPCOM000202411D
Publication Date: 2010-Dec-15
Document File: 3 page(s) / 101K

Publishing Venue

The IP.com Prior Art Database


Disclosed is an architecture to mine the interesting keywords from the SMS data residing on a set of mobile devices. The mining is done using the computational capabilities of the mobile phones in combination with servers in a privacy preserving fashion. The mined keywords can be used to build interesting services like targeted advertisement, recommendation systems etc.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 54% of the total text.

Page 01 of 3

System and architecture for Mining SMS Messages on Mobile Devices Preserving Users' Privacy


The step by step procedure for mining the SMS data across a set of mobile phone users is discussed below:

Step 1: A set of mobile devices Sm is considered. Each device in Sm has a set of SMS messages residing locally. In order to mine this SMS data residing on these devices, each device reads its local SMS data and creates a list ln containing most frequent keywords along with its frequency. While creating this list, some of the common keywords like 'and', 'the' etc. may be filtered out from consideration and a suitable predicate (e.g. at least 'k' occurrence) may be used to define which keywords are frequent. This step is done locally on each mobile device using the computation and memory resources of the mobile device.

Step 2:A translation server T---s is considered which can handle at least following set of requests from the mobile devices:

1. Take keyword (encoded or native) list as input and provide encoded list as output. The encoding may be done by replacing each of the keywords with uniquely encoded identifiers.

2. Take two encoded keyword list as input, merge them and provide an encoded list as output. This merging ensures that frequency of each of the keywords in the output list is the sum of the frequency in the input list.

3. Take an encoded word and return the decoded word. This operation is successful only if the server requesting the decoding is same as the one who requested encoding earlier.

In this step, each of the mobile devices sends their local native (i.e. non-encoded) list of most frequent keywords ln to Ts and receives back the encoded list le from Ts .

Step 3:

The aggregation of the individual encoded list le residing on the individual mobile devices is done using a tree like server architecture Gtas shown in Figure. The leaves of this architecture contain individual mobile devices. In this step, each of the mo...