Demos & Panels



Time to Rethink Database Teaching in India?
Panelists: Prof. Volker Markl (TU-Berlin), Prof. Sudarshan (IIT B), Dr. Arnab Bhattacharya (IIT K), Prof. Srinath Srinvasa (IIIT B), Sanjeev Kumar (Informatica)
Moderator: Srikanta Bedathur (IIIT-Delhi)

In the last decade or so, the scope of database research has witnessed an explosive expansion.  When one looks at the research publications in top DB conferences, it is not surprising to see papers on topics ranging from machine learning to distributed systems, multi-modal datasets to petabytes of scientific data, solutions customized for modern hardware to visualization-driven analytics, and so on. In fact, these papers dominate the proceedings compared to papers on "traditional" DB topics.

Given this, an important question that the database research community faces is: are we teaching the right things in our database courses? There have been many panels to discuss these issues (including a recent panel in SIGMOD). The typical conclusion has been: the fundamentals of database systems are important, and they have to be taught. Unfortunately, this means very little time can be devoted to teaching the "hot" topics to undergrads and graduate students, resulting in little or no preparation for state of the art database methods and building of data-centric applications.

But now, there is new twist which could disrupt this traditional thinking: the easy availability of online courses. Technology and the popularity of online classes are radically transforming the educational process on both national and international scales. Nothing exemplifies this better than the great amount of excitement that the online versions of AI and ML courses of Stanford University have generated.  With this new wave of teaching technology and methodology on our hands, we are forced to revisit our presumptions about pedagogy -- particularly in the fast broadening field like databases, and in a region like India, where access to network is easier than good teachers!

The questions we explore in this panel include: whether students be expected to learn the fundamental courses only through online courses, while lecture-hall courses be limited to more advanced topics? Or the model should be inverted? How can Industry actively participate in such teaching method so that they can get the best trained data-centric graduates?

Invited Academic Demos

Dynamic Question Paper Template Generation using GA Approach
Dimple V Paul Shankar B. Naik, Priyanka Rane and Jyoti D. Pawar


Automatic Generation Of Question Paper From Question Bank (AGQPQB) is a tool that records huge amount of data related to different undergraduate and postgraduate degree courses and their subjects offered at a University along with the syllabus details, instructor details, examination details and question bank details. Details related to the examination include the examination type (for example In-Semester Examination, End-Semester Examinations etc,),the taxonomy type (such as Blooms, Solo etc.,), the respective levels of the taxonomy (like Knowledge, Understanding etc.,) and their corresponding verbs (such as Describe, Explain, Write etc,). The question bank details include details of a question and question type (for example Long Answer, Very Short Answer, Multiple Choice Questions etc). 
Using the above data, AGQPQB   generates question paper templates. Instructor input for a template is: 1) levels of an educational taxonomy 2) topics of the syllabus of a subject. Generated templates help to maintain the quality of a question paper by considering proper weightage across taxonomy levels; proper syllabus coverage across topics, appropriate time duration and different difficulty levels such as Low, Medium, High etc., .The difficulty level of a question paper depends on the weightage assigned to different levels of the taxonomy. For example a question paper template of High difficulty level which uses Blooms Taxonomy will have more weightage for Synthesis and Evaluation levels as compared to Knowledge, Understanding, Application and Analysis levels.  
Main modules of AGQPQB are Course Registration, Subject Registration, Taxonomy Registration, Examination Registration, Instructor Registration, Template Generation, Question Bank Module and Question Paper Generation.


eAgromet: An ICT-based agro-Meteorological Advisory System

P.Krishna Reddy, B.Bhaskar Reddy,  P.Gowtham Srinivas, D.Satheesh Kumar, D.Raji Reddy, G.Sreenivas, L.S. Rathore, K.K. Singh, N.Chattopadhyay
International Institute of Information Technology Hyderabad, Acharya NG Ranga Agricultural University - Hyderabad,  Indian Meteorological Department


Based on the weather forecast, Indian meteorological department (IMD)  is  disseminating  agromet advisory bulletins to farmers and stakeholders. The  agromet bulletins  contain possible agricultural risk mitigation steps.   The agromet advisory bulletins are prepared by  about 130 agro  meteorological management functional units (AMFUs) spread allover India. Based on weekly weather forecast,  AMFUs prepare district-level agromet bulletins.  The eAgromet is an ICT-enabled agro-meteorological advisory system being developed in India by IIIT, Hyderabad, Acharya NG Ranga Agricultural University and IMD.  The objective is to build a system to improve the efficiency of agromet advisory bulletin preparation and dissemination process by exploiting information and communication technologies. 



DBridge: Holistic Optimization of Database Applications
Karthik Ramachandra (IIT Bombay), Ravindra Guravannavar (IIT Hyderabad), Mahendra Chavan (Sybase), Prabhas Samanta (Sybase),  Sudarshan (IIT Bombay).


We present DBridge, a novel static analysis and program transformation tool to optimize database access. Traditionally, rewrite of queries and programs are done independently, by the database query optimizer and the language compiler respectively, leaving out many optimization opportunities. Our tool aims to bridge this gap by performing holistic transformations, which may include both program and query rewrite.
Many applications invoke database queries multiple times with different parameter values. Such query invocations made using imperative loops are often
the cause of poor performance due to random I/O and round trip delays.  In practice, such performance issues are addressed by manually rewriting the
application either to make the query set oriented, or by asynchronously issuing queries to overlap query executions. Such manual rewriting of programs is often time consuming and error prone.  Guravannavar et. al.(VLDB 2008) propose program analysis and transformation techniques for automatically rewriting an application to make it set oriented.  Chavan et. al (ICDE 2011) propose techniques for asynchronous query submission. DBridge implements these program transformation techniques for Java programs that use JDBC to access  the database. 
In this demonstration, we showcase the holistic program/query transformations that DBridge can perform, over a variety of scenarios taken from real-world applications. We then walk through the design of DBridge, which uses the SOOT optimization framework for static analysis. Finally, we 
demonstrate the performance gains achieved through the transformations.

(Demonstrated at ICDE 2011)


The Picasso Database Query Optimizer Visualizer
Institution:  Database Systems Lab, Indian Institute of Science
Presenter: Bruhathi H S


In this demo, we present the Picasso visualization tool, which provides powerful visual metaphors to profile and analyze the intriguing behavior of modern database query optimizers. The Picasso diagrams, which often appear similar to cubist paintings, provide a variety of interesting insights, including that current optimizers make extremely fine-grained plan choices; that the plan optimality regions may have highly intricate patterns and irregular boundaries, indicating strongly non-linear cost models; that non-monotonic cost behaviors exist where increasing result cardinalities decrease the estimated costs; and, that the basic assumptions underlying the research literature on parametric query optimization often do not hold in practice.

Picasso has been employed as (i) a query optimizer analysis, debugging, and redesign aid by system developers; (ii) a query optimization test-bed by database researchers; and (iii) a query optimizer pedagogical support by database instructors and students. A particularly useful application is the efficient identification of robust plans that are resistant to selectivity estimation errors.

(Demonstrated at VLDB 2010, Singapore.)


CODD: Construction of Dataless Databases
Institution:  Database Systems Lab, Indian Institute of Science
Presenter: I Nilavalagan


A fundamental requirement in the effective design and testing of database engines and applications is the ability to easily construct alternative scenarios with regard to the database contents.  A limiting factor, however, is that the time and space overheads incurred in creating and maintaining these databases may render it infeasible to model the desired scenarios.  In this demo, we present CODD, a graphical tool that attempts to alleviate these difficulties through the construction of "dataless databases". Specifically, CODD implements a visual interface through which databases with the desired meta-data characteristics can be efficiently simulated without persistently generating and/or storing their contents. Apart from the dataless modes, it also provides potent additional functionalities, including both size-based and cost-based scaling of metadata instances, automatic porting of meta-data across database engines, and identification of hidden constraints in legacy code.


Institution: International Institute of Information Technology, Bangalore
Presenters: Prerna Atri and Shreya Barsaiyan.


RootSet  is  a  web  based  portal  for  representing  operational  knowledge pertaining  to  the  real world.  The  project's  knowledge  database  is  created  through  crowd-sourcing  of  formalized knowledge.  It  can  be  used  by  communities  for developing  a  schema  dynamically  for  a  specific domain through knowledge aggregation. 
 In  RootSet,  the  knowledge  base  is  a  collection  of  abstract  objects represented  as  concepts.  A mechanism to establish different relationships between concepts is provided. However, there are two basic semantic relationships which are necessary for a concept to be represented: the “IS-A” relationship  that  leads  to  the  concept  hierarchy  and  the  “IS-IN”  relationship  that  leads  to  the  containment hierarchy.  Every  user  can  use  the  existing  concepts  and/or  create  his/her  own concepts and conceptual hierarchies. The  demonstration  will  consist  of  a  modeling example  using  the  RootSet  portal  to  explain  the functioning of the same.


Contextual Knowledge Based Assistance for Academics (CKAA)
Institution: International Institute of Information Technology, Bangalore
Presenters: Rohit Jain and Vibhor Jain


Seekha is a system for academics that cater to knowledge requirements of its various users such as students, researchers, academic committees, and institutions. The system then contextually examines, correlates and aggregates information based partly on domain knowledge and partly on observing co-occurrence patterns. The system evolves continuously along with the user, by learning his/her interests, and by involving in the continuous process of knowledge extraction, knowledge aggregation and knowledge dissemination. The system eventually serves as an Intelligent Knowledge backbone catering to the specific requirements of users. Seekha will act as a hub of resources for students, researchers, teachers, etc. One of the objectives of Seekha is to connect all the institutes of India so that it can consolidate academic activities, events and interests in the technical education area across the country. It’s a Search Engine which will be using concept extraction and contextual data retrieval to gather information.