Saumya Pandey

Developed and implemented a sentiment analysis system that evaluates Facebook statuses for sentiments and states like anniversary,achievements,emotions etc and generates appropriate greetings in response. This is achieved by building different sentiments groups and buckets using specialized dictionaries that are derived from the help of topical modeling on data. Once the data is grouped into appropriate buckets, the response for each status is obtained. Utilized tools like mallet, SentiwordNet,… Show more

Developed and implemented a sentiment analysis system that evaluates Facebook statuses for sentiments and states like anniversary,achievements,emotions etc and generates appropriate greetings in response. This is achieved by building different sentiments groups and buckets using specialized dictionaries that are derived from the help of topical modeling on data. Once the data is grouped into appropriate buckets, the response for each status is obtained. Utilized tools like mallet, SentiwordNet, Java programming language and a few standard NLP packages. Show less

Performed topic modelling on data archives of BBC news articles to categorize them efficiently using the Mallet modelling software. Also performed modelling on data set containing reviews about the top american city,to highlight the topics or factors that mostly get mentioned when people talk about a livable city.
Utilized the results to report the findings of document categorization and topical analysis.

1)Utilized standard NLP tools to parse English novels and identify relations of type speak(x,y)( X speaks to Y ). The data was divided into development set and testing set. The development set once parsed, was observed for annotators and fields like lemma,ner,dcoref etc to identify patterns that indicate relation speak(x,y). These patterns are then encoded and the testing set is used to get results based on identified patters.

2) Utilized the BerkelyLm language model and Soundex libaray… Show more

1)Utilized standard NLP tools to parse English novels and identify relations of type speak(x,y)( X speaks to Y ). The data was divided into development set and testing set. The development set once parsed, was observed for annotators and fields like lemma,ner,dcoref etc to identify patterns that indicate relation speak(x,y). These patterns are then encoded and the testing set is used to get results based on identified patters.

2) Utilized the BerkelyLm language model and Soundex libaray to train the LM,identify spelling errors in the testing text and generate suggestions for correction . Show less

Client: Thomson Reuters

Project aims to analyze publication outputs of 3 infectious diseases (Ebola, Swine Influenza and Avian Influenza) for the time period 1996 to 2014 by conducting research on scholarly publications in order to provide quantitative analysis & data visualization to gain insights.

Project involves the following tasks:
a) Performing trend analysis on the outbreak of diseases and research inputs in order to investigate correlations between the outbreak of… Show more

Client: Thomson Reuters

Project aims to analyze publication outputs of 3 infectious diseases (Ebola, Swine Influenza and Avian Influenza) for the time period 1996 to 2014 by conducting research on scholarly publications in order to provide quantitative analysis & data visualization to gain insights.

Project involves the following tasks:
a) Performing trend analysis on the outbreak of diseases and research inputs in order to investigate correlations between the outbreak of diseases and publication outputs.
b) Identifying collaborations between researchers within and across organizations on ebola, avian influenza and swine influenza topics.
c) Identifying disciplines that collaborated the most and the percentage of research activities that are multidisciplinary.
d) Formulating recommendations on researchers/organizations for new researchers based on their research interests through the generated visualization.

Intend to use Bipartite Graph, Network Graph and Line Graph Visualizations as well as Burst Analysis for the aforementioned tasks.

(Team size: 5)

[ MySQL, Tableau 8.3, Gephi 0.8.2, Navicat 11.1.10, Sci2 1.1 and Inkscape 0.91 ]
Show less

1)Built a simple reflex agent that behaves intelligently to perform desired work under defined scenarios.

2) Designed and implemented a python code that guides a robot arm in a robot world to search blocks and move them to the goal locations.

3) Designed a game of checkers in python language using advanced search algorithm of minmax with alpha beta pruning.Implemented efficient evaluation functions to make the code more intelligent.

Developed all the essential future_allocation, future_free, future_get and future_set system calls required for the implementation of futures in Xinu Operating System on Raspberry Pi. It involved synchronizing the asynchronously produced data, thereby potentially improving the ability to manage concurrent activities far more efficiently. Implemented a test program to simulate our futures implementation on a sample Producers-Consumers problem. (As a part of Advanced Operating Systems… Show more

Developed all the essential future_allocation, future_free, future_get and future_set system calls required for the implementation of futures in Xinu Operating System on Raspberry Pi. It involved synchronizing the asynchronously produced data, thereby potentially improving the ability to manage concurrent activities far more efficiently. Implemented a test program to simulate our futures implementation on a sample Producers-Consumers problem. (As a part of Advanced Operating Systems Course-work, Team size: 2) Show less

For cloud computing course I worked on Hadoop MapReduce programs for below three applications.

1. Hadoop Word Count
Hadoop Word Count is a program which counts the number of occurrences of each word in a given text input dataset. It uses Map program to map each word in input file with count 1. Reducer is then used to combine and add all counts for each word to display total word count. Combiner is used as added feature so as to reduce extra overhead of reducer.

2. Hadoop Page… Show more

For cloud computing course I worked on Hadoop MapReduce programs for below three applications.

1. Hadoop Word Count
Hadoop Word Count is a program which counts the number of occurrences of each word in a given text input dataset. It uses Map program to map each word in input file with count 1. Reducer is then used to combine and add all counts for each word to display total word count. Combiner is used as added feature so as to reduce extra overhead of reducer.

2. Hadoop Page Rank
A parallel version of PageRank using the programming interfaces of the Hadoop MapReduce framework.

3. Hadoop Blast
Hadoop-Blast is an advanced Hadoop program which helps BLAST, a bioinformatics application, to utilize the computing capability of Hadoop. Map reduce framework is used to implement Hadoop Blast.

Show less

Developed a simple in-memory file-system for Xinu on Raspberry Pi using inode and various file table structures to handle data blocks. Developed standard file system functions such as fcreate, fopen, fread, fwrite, fseek, fclose etc. The project implementation helped to gain further insights on inode structures, bit manipulations , block allocation and system calls in file systems. (As a part of Advanced Operating Systems Course-work, Team size: 2)

Implemented a user interactive code to setup 2D and 3D geometry, add projections,illumination and animation using WebGl. The application code was implemented using Javascript and HTML.Operations like transformations and projections were all user controlled using events and other graphical operations like adding illumination and animation into scene was successfully implemented.(As a part of Advanced Computer Graphics coursework)

Implemented a shared memory IPC mechanism in Xinu on Raspberry Pi similar to the one provided by System V using shell commands. This mechanism of IPC was achieved by allowing a process to create/reserve a shared memory segment. Another set of processes can then write to this region directly. Developed the essential shmget (Get), shmat (Attach) and shmdt (Detach) system calls for interacting with shared memory. Multiple invocations of its shell command would create a shared memory segment, write… Show more

Implemented a shared memory IPC mechanism in Xinu on Raspberry Pi similar to the one provided by System V using shell commands. This mechanism of IPC was achieved by allowing a process to create/reserve a shared memory segment. Another set of processes can then write to this region directly. Developed the essential shmget (Get), shmat (Attach) and shmdt (Detach) system calls for interacting with shared memory. Multiple invocations of its shell command would create a shared memory segment, write to a pre-reserved shared memory segment, and display the contents of the shared memory. The kernel bookkeeps each shared memory in a data structure. (As a part of Advanced Operating Systems Course-work, Team size: 2) Show less

Developed a multi-threaded program on Linux for repeated matrix multiplications using POSIX Threads where the output of the computation depends on user’s input. The program provides functionality to pause, resume, restart, stop and display the intermediate values of matrix multiplication at a particular instant.The program simultaneously monitors the terminal window as well as the content of a shared file to read user’s input. The user may provide input though the terminal window or a text… Show more

Developed a multi-threaded program on Linux for repeated matrix multiplications using POSIX Threads where the output of the computation depends on user’s input. The program provides functionality to pause, resume, restart, stop and display the intermediate values of matrix multiplication at a particular instant.The program simultaneously monitors the terminal window as well as the content of a shared file to read user’s input. The user may provide input though the terminal window or a text editor. (As a part of Advanced Operating Systems Course-work, Team size: 2)
Show less

Implemented basic features of Reverse Polish Notation (RPN) Calculator. It performs addition, subtraction, multiplication, division and modulo operations on whole numbers (i.e. octal, hexadecimal or binary), decimal fractions and floating point numbers.

The program reads numbers and operators via Non-Canonical input processing mode where the tokens are separated by white space. The implementation of pushing and popping of operands & operators was achieved using stack data structure. The… Show more

Implemented basic features of Reverse Polish Notation (RPN) Calculator. It performs addition, subtraction, multiplication, division and modulo operations on whole numbers (i.e. octal, hexadecimal or binary), decimal fractions and floating point numbers.

The program reads numbers and operators via Non-Canonical input processing mode where the tokens are separated by white space. The implementation of pushing and popping of operands & operators was achieved using stack data structure. The program reads input tokens and upon encountering a number, the program places the number on the stack. When the program reads an operator, it applies the operator to the two numbers at the top of the stack, puts the result on the top of the stack and prints the result with the fewest required digits.The program was also able to gracefully handle errors such as divide by zero and improperly formatted numbers. (As a part of Advanced Operating Systems Course-work, Team size: 2)
Show less

Designed a framework for linking multiple heterogeneous Information systems via a centralized server using MatLab, SQL Server & Visual Studio. The framework aimed at automating the process of recording and retrieving information in public offices to assist people with disabilities while requesting their services. It facilitates the biometric process of minutiae based finger print matching for the purpose of user identification and authentication. (Team size: 2)