What is the difference between short and long read?
The size, short reads can be considered when you have sequences < 200-400 bp (Illumina, SOLiD, current IonTorrent) and long when your sequences is > 400 bp (Sanger, 454, PacBio, future IonTorrent and NanoPore)
What are the comparative advantages of each?
Short reads are cheaper to produce (less reagents and cycles) and are completely adequate when you have a reference genome but longer reads provides more sequences that can be used to assemble a genome or transcriptome.
Which one of them is used now-a-days for sequence assembly?
Both are used, you can assemble a small genome with short or long reads, you can mix short and long reads to improve any assembly, but because costs, short reads are most commonly used.
Which one is preferred for future from both the computational and analytical point of view?
Longer reads, because technology is leading to cheaper long reads.
Nice.