You are on page 1of 5

AMITY UNIVERSITY

Uttar Pradesh Lucknow Campus


Amity Institute of Information Technology
WEEKLY PROGRESS REPORT
For the week commencing from: 12 June 2023 – 18 June 2023

WPR No. 2

Name of the student: Shaurya Upadhyay

Enrollment Number: A7304822105

Program: Bachelor of Computer Applications (2019-2022) Semester: II

Name of the Non-Teaching Credit Course: Summer Internship – 1 [ETTP100]

Organization Name: AIIT, Amity University Uttar Pradesh, Lucknow

Faculty Guide’s Name: Dr. Ajay Pratap

Project Title: Detecting Fraud Apps Using Sentiment Analysis Using Machine Learning

Targets set for the week Achievements for the week Future work plans
.
Here is a breakdown of Here are suggested Here are suggested future
suggested targets for each day achievements for each day of work plans for each day of the
of the week in the project the week in the project week in the project "Detecting
"Detecting Fraud Apps Using "Detecting Fraud Apps Using Fraud Apps Using Sentiment
Sentiment Analysis Using Sentiment Analysis Using Analysis Using Machine
Machine Learning": Machine Learning": Monday: Learning": Monday: 1.
1. Conducted a comprehensive Explore advanced sentiment
Monday: literature review on fraud analysis techniques, such as
1. Collect and preprocess a detection in mobile apps, deep learning models or
substantial dataset of app sentiment analysis techniques, transformer-based models, to
reviews, ensuring data quality and machine learning improve the accuracy and
and addressing any missing algorithms. 2. Defined the granularity of sentiment
values or inconsistencies. project's objectives, scope, analysis. 2. Investigate the use
2. Perform exploratory data and success criteria. 3. Set up of domain-specific sentiment
analysis to gain insights into the development environment, lexicons or pre-trained
the distribution of app including necessary libraries language models tailored for
reviews, sentiment, and and frameworks. 4. Gathered a app reviews to capture
potential fraud patterns. diverse dataset of app reviews, nuances and domain-specific
3. Implement a baseline both language. 3. Experiment with
sentiment analysis model different text preprocessing
using a machine learning techniques, such as stemming,
algorithm, such as Naive lemmatization, or part-of-
Bayes or logistic regression, speech tagging, to further
and evaluate its performance. enhance the quality of the text
4. Set up the project data. 4. Consider
repository, organize the incorporating semantic
project structure, and define analysis or entity recognition
the necessary dependencies. to extract additional
information from the app
Tuesday: reviews. Tuesday: 1.
1. Research and experiment Research and implement
with advanced sentiment advanced feature engineering
analysis techniques, such as techniques, such as word
deep learning models (e.g., embeddings (e.g., Word2Vec,
LSTM, CNN) or transformer- GloVe) or contextualized
based models (e.g., BERT, embeddings (e.g., BERT,
GPT), to improve sentiment ELMo), to capture richer
analysis accuracy. semantic information in the
2. Explore methods for app reviews. 2. Explore the
addressing the challenges of use of topic modeling
sentiment analysis in app algorithms, such as Latent
reviews, such as handling Dirichlet Allocation (LDA) or
domain-specific language, Non-Negative Matrix
sarcasm, or context ambiguity. Factorization (NMF), to
3. Train and fine-tune the identify key topics or themes
advanced sentiment analysis in the app reviews related to
model using a labeled dataset, fraud. 3. Investigate the
and evaluate its performance potential impact of temporal
against the baseline model. features, such as review
4. Document the timestamps or release dates,
experimentation process, on fraud detection and
including model architectures, incorporate them into the
hyperparameters, and feature extraction pipeline. 4.
performance results. Consider feature engineering
techniques specific to fraud
Wednesday: detection, such as extracting
1. Investigate fraud detection linguistic patterns or syntactic
techniques that can be applied structures indicative of
to app reviews, considering fraudulent reviews.
approaches like anomaly Wednesday: 1. Experiment
detection, supervised with different machine
classification, or unsupervised learning algorithms, such as
clustering. gradient boosting machines,
2. Implement a fraud detection deep neural networks, or
model using machine learning hybrid models, to improve the
algorithms like random fraud detection performance.
forests, support vector 2. Explore techniques to
machines, or ensemble address model interpretability,
methods, and evaluate its such as SHAP values, LIME,
performance. or rule-based approaches, to
3. Explore techniques to provide explanations for the
handle class imbalance in the predictions made by the
dataset, such as oversampling, machine learning models. 3.
undersampling, or generating Investigate the use of active
synthetic samples. learning strategies to
4. Compare and analyze the iteratively label and
results of the fraud detection incorporate additional high-
model against the baseline impact data points into the
sentiment analysis model, training process, improving
identifying strengths and the model's performance over
weaknesses. time. 4. Consider
incorporating external data
Thursday: sources, such as social media
1. Research and implement feeds or user behavior data, to
techniques to enhance fraud enrich the fraud detection
detection using contextual models and capture additional
information, such as user signals. Thursday: 1.
behavior, app metadata, or Experiment with ensemble
review timestamps. learning techniques, such as
2. Evaluate the impact of stacking, blending, or model
incorporating additional averaging, to combine the
features, such as user ratings, predictions of multiple
app category, or reviewer machine learning models and
reputation, on fraud detection further boost fraud detection
performance. performance. 2. Investigate
3. Experiment with feature techniques for model
selection or dimensionality calibration to ensure well-
reduction techniques to calibrated probability
improve model efficiency and estimates, allowing for more
reduce overfitting. reliable decision-making. 3.
4. Conduct a thorough Explore transfer learning
evaluation of the fraud approaches, where knowledge
detection model, considering learned from related tasks or
performance metrics like domains can be leveraged to
accuracy, precision, recall, F1- improve fraud detection in app
score, and area under the ROC reviews. 4. Consider the
curve. potential impact of adversarial
attacks and explore techniques
Friday: to enhance the robustness of
1. Investigate methods to the fraud detection models
interpret and explain the against such attacks. Friday:
predictions of the fraud 1. Evaluate the scalability and
detection model, such as efficiency of the fraud
feature importance analysis or detection system, considering
model-agnostic techniques. factors such as computational
2. Assess the robustness and resources, response time, and
generalizability of the fraud memory requirements. 2.
detection model by Investigate the potential for
conducting cross-validation or deploying the fraud detection
testing on external datasets. system in a real-time or
3. Explore techniques to streaming environment,
mitigate biases in the fraud enabling continuous
detection model, ensuring fair monitoring of app reviews for
and unbiased predictions fraud detection. 3. Conduct
across different user groups or rigorous testing and validation
app categories. of the entire system,
4. Document the model considering various scenarios,
evaluation results, including edge cases, and real-world
performance metrics, data. 4. Perform a comparative
interpretability analysis, and analysis of the developed
discussions on biases and system against existing fraud
limitations. detection methods or
commercial solutions,
Saturday: highlighting the strengths and
1. Prepare a comprehensive limitations of the proposed
project report summarizing approach. Saturday: 1.
the methodology, findings, Prepare a detailed
and conclusions of the project. documentation of the project,
2. Create visualizations and including the implemented
presentations to effectively methodologies, findings,
communicate the results and challenges faced, and lessons
insights gained from the fraud learned. 2. Reflect on the
detection model. limitations and potential
3. Reflect on the project's ethical considerations of the
limitations, challenges faced, developed fraud detection
and potential future directions system, addressing issues such
for improving the fraud as bias, privacy, or fairness. 3.
detection system. Explore possibilities for
4. Conduct a final review of collaboration or further
the project deliverables, research in the field of fraud
ensuring all code, detection in mobile apps,
documentation, and sentiment analysis, or related
supplementary materials are areas. 4. Present the project's
organized and complete. outcomes and insights to
stakeholders, such as the
Remember to adjust the project team, advisors, or
targets based on the specific potential users, and gather
requirements and timeline of feedback for future
your project. Regularly improvements. Remember to
communicate with your team adjust
members, seek guidance from
mentors or advisors, and track
progress to ensure successful
completion of the project.

Signature of the Student: ___Shaurya Upadhyay____________________


(Name of Student)

Signature of the Faculty Guide: __________________


(Name of Guide)

You might also like