Abstract:
Digital Forensic Investigation (DFI) is the investigation of crimes that involve the
investigation of digital evidence, data and communication that are carried out on the
suspectâAZs computers. DFI has become a research trend in the field of data mining
because crimes ratio that carried out through computers is increasing. Moreover the
essence of data mining in DFI is getting important because the capacity of computer
storage and consequently size of data in computers are increasing with the passage of time.
It becomes difficult to take the manual investigation of a computerâAZs data because
it consumes too much time.In existing systems, data on crime scenes are retrieved from
computers and clustered using clustering techniques on the basis of subjects defined by an
investigator. The subjects are sensitive words related to the crimes. These clusters help
in identifying relevant data on specific subjects which are useful for further investigation.
The approach is also known as subject-based semantic document clustering.A drawback
of the approach is that these generated clusters are concentrated on subjects, provided
by the investigator and not on the subjects found from the suspectâAZs computer.
In this research we have also applied subject-based semantic clustering on documents
found in the suspectâAZs computer. In order to resolve the above mentioned issue, the
proposed approach first analyses documents and recommends subjects to the investigator
for his selection. Then the investigator provides subjects for clustering of documents. The
proposed approach applies overlapping clusters on the provided subjects and generates
another generic cluster of documents that do not fall in the clusters of provided subjects.
In addition, the generic cluster can be further passed to another cycle of this process
for additional investigation. The experimental results show that the proposed approach
provides comparatively more accuracy and flexibility than the existing systems.
