First Progress Report

ON

FileFlow
Submitted in partial fulfilment of the requirements for the award of the degree

of

BACHELOR OF TECHNOLOGY

in

COMPUTER SCIENCE AND ENGINEERING

by

Aishita Saini Anshita Chaudhary Drishti Bansal

01015002720 02415002720 05015002720

Guided by

Ms. KIRTI DAHIYA

Assistant Professor

DEPARTMENT OF COMPUTER SCIENCE & ENGINEERING

MAHARAJA SURAJMAL INSTITUTE OF TECHNOLOGY

(AFFILIATED TO GURU GOBIND SINGH INDRAPRASTHA UNIVERSITY,

DELHI)

DELHI – 110058

2020-2024
 TABLE OF CONTENTS

Abstract …………………………………………………………………………...… 1

Introduction ………………………………………………………………………… 2

Literature Survey ……………………………………………………………...…… 3

Objective ……………………………………………………………………….…… 4

Methodology of study …………………………...………………………………….5

Limitations …………………………………………………………………………...6

References ………………………………………………...………………………….7
 ABSTRACT

In the realm of Artificial Intelligence, our project, FileFlow aims to introduce a Large
Language Model (LLM) based system designed to enhance conversational AI. While
our primary focus is on enabling meaningful interactions, we also aim to integrate the
capability to navigate through a diverse range of document formats. This synthesis
focuses on providing users with a versatile AI assistant that caters to both casual
conversations and document-related tasks.

This report delves into the cutting-edge technologies driving a paradigm shift in
information interaction, transcending traditional document management approaches.
Leveraging advanced natural language processing (NLP) algorithms, users can engage
in dynamic conversations with their files, extracting insights, asking questions, and
gaining a deeper understanding of their content.

Key to this innovation is the integration of AI-driven chat companions, augmenting

the user experience with personalized assistance and fostering interactive dialogue.
This convergence of technology facilitates a fluid exchange of information, blurring
the lines between human interaction and machine intelligence.
 INTRODUCTION

Our project aims to develop a user-friendly tool called FileFlow, designed to facilitate
easy communication and comprehension of various file formats such as documents,
presentations, spreadsheets, and more. This project will serve as a conversational
platform, allowing individuals to interact with their files as if they were chatting with
a knowledgeable friend.

The primary objective is to create a seamless experience where users can engage in
natural conversations with their documents, asking questions, seeking clarification,
and extracting relevant information effortlessly. We want to eliminate the complexity
typically associated with navigating and understanding diverse file types, making it
accessible to anyone, regardless of technical expertise.

Additionally, this project will incorporate features that promote casual conversation
and learning, including the ability to chat with an AI companion. This AI friend will
offer assistance, provide explanations, and engage users in friendly discussions,
making the process of working with documents both enjoyable and educational.

By achieving these objectives, we aim to empower individuals to harness the full

potential of their files, fostering productivity, creativity, and collaboration in various
personal, academic, and professional settings. FileFlow will democratize access to
information, enabling users to communicate with their documents in a way that feels
natural and intuitive, ultimately enhancing their overall experience and satisfaction.
 LITERATURE SURVEY

Natural Language Processing is an area of research within Artificial Intelligence (AI)

that is concerned with giving computers the ability to understand natural language
(spoken and written) in the same way a human could [1]. Knowledge of
computational linguistics (rule-based modelling of human language), statistics,
machine learning and deep learning are used either individually or combined to
achieve the aforementioned goal [2]. Few of the researched tasks of NLP are
Automatic Summarization (Automatic summarization produces an understandable
summary of a set of text and provides summaries or detailed information of text of a
known type), Co-Reference Resolution (Co-reference resolution refers to a sentence
or larger set of text that determines all words which refer to the same object),
Discourse Analysis (Discourse analysis refers to the task of identifying the discourse
structure of connected text i.e. the study of text in relation to social context)[3]. Many
researchers worked on NLP, building tools and systems which makes NLP what it is
today. Tools like Sentiment Analyser, Parts of Speech (POS) Taggers, Chunking,
Named Entity Recognitions (NER), Emotion detection, Semantic Role Labeling have
a huge contribution made to NLP.

A chatbot is an artificially intelligent creature which can converse with humans. This
could be text-based, or a spoken conversation (in case of voice-based queries). Chat
bots are basically used for information acquisition. It can run on the local PCs and
mobile phones, though most of the time it is accessed through the internet. It is a
conversational agent which interacts with users in a certain domain or on a particular
topic with input in natural language sentences. Mainly a chatbot works by a user
asking some question or initiating a new topic of discussion.[5] a paper by Tarun
Lalwani, Shashank Bhalotia conveyed Chat bots can be referred as software agents
that pretend as human entity. These are the agents with AI embedded and using NLP
they can answer to user questions. Predefined knowledge base helps develop a
response to the query [6].
Chatbot design [7] typically relies on parsing techniques, pattern matching strategies
and Natural Language Processing (NLP) to represent the chatbot knowledge. The
latter is the dominant technique thanks to the popularization of libraries and cloud-
based services such as DialogFlow [8] or IBM Watson Assistant [9], which rely on
Machine 4 Learning (ML) techniques to understand the user input (based on a set of
training sentences provided as part of the chatbot definition) and provide user-friendly
interfaces to design the conversational flow.

Document structure analysis can be regarded as a syntactic analysis problem. The

order and containment relations among the physical or logical components of a
document page can be described by an ordered tree structure and can be modelled by
a tree grammar which describes the page at the component level in terms of regions or
blocks.[10] Document structure analysis can be regarded as a syntactic analysis
problem. The order and containment relations among the physical or logical
components of a document page can be described by an ordered tree structure and can
be modelled by a tree grammar which describes the page at the component level in
terms of regions or blocks. [11]. The global chatbot market is projected to reach 2
billion dollars by 2024, growing at a CAGR (compound annual growth rate) of
29.7%.2 This widespread interest and demand for chatbot applications has
emphasized the need to be able to quickly build complex chatbot applications
supporting AI-based natural language processing
 OBJECTIVE

Our project seeks to redefine the conventional scope of conversational AI by

combining linguistic understanding with the ability to interact with various file
formats. The goal is to offer users a well-rounded tool that seamlessly transitions
between casual discussions and document-based queries. The primary objectives of
this project are:

1. User-Friendly Interface: Design an intuitive and user-friendly interface that

facilitates easy interactions, document uploads, and retrieval of information, ensuring
a seamless user experience.

2. Seamlessly Blending Analysis with Casual Chat: Beyond simply understanding

information, this project strives to mimic the natural flow of human conversation. We
envision a model that seamlessly transitions from in-depth file analysis to engaging
dialogue, mirroring the way we interact with other AI chatbots. This innovative
feature removes the coldness of traditional analysis tools, offering a more natural and
enjoyable way to interact with information.

3.Optimize Natural Language Understanding in Document Queries: Fine-tune the

Local Language Model (LLM) using advanced Natural Language Processing (NLP)
techniques to enhance its understanding and responsiveness specifically tailored for
document-related queries.

4. User Engagement and Satisfaction: Focus on user engagement metrics and

feedback to continually refine and improve the AI assistant, ensuring high levels of
user satisfaction with the conversational and document interaction aspects.

This project isn't about grandiosity but rather about creating a tool that seamlessly
integrates into users' workflows, providing practical utility in both casual
conversations and document-related tasks.
 METHODOLOGY OF STUDY

1.Project Inception and Planning: Clearly outline project's goals, focusing on

enhancing document interaction through AI and supporting casual conversations.
Identify the range of document formats (PDF, DOCX, PPTX, etc.) the system will
support and define the conversational capabilities of the AI. Allocate personnel
specialized in AI, NLP , UI/UX design, and software development.

2. Requirements Gathering: Define the desired user experience, focusing on natural

interaction and efficient document handling. Set targets for the AI's understanding of
user queries and t he accuracy of document parsing and information retrieval. List
features such as querying document content, summarizing documents, answering
context-specific questions, and facilitating engaging AI conversations.

3. Technology Stack Selection: Choose AI frameworks and NLP libraries for

processing and understanding natural language. Select programming languages and
frameworks that support rapid development and integration of AI functionalities.
Decide on cloud platforms for deploying the application, considering scalability and
AI service offerings.

4. Develop the AI component capable of understanding and responding to user queries

about documents and engaging in casual conversation. Enable the AI to navigate,
parse, and extract information from various document formats.

5. Frontend and User Interface Development: Create a user-friendly web interface

that facilitates easy interaction with the AI for document processing and casual
conversations.

6. Integration and Testing: Seamlessly integrate the frontend, conversational AI,

and document processing backend. Ensure the system meets usability, performance,
and accuracy goals.
 LIMITATIONS

1. Natural Language Processing Accuracy: While the project aims to enable natural
conversations with documents, the accuracy of understanding and responding to user
queries in natural language is a significant challenge. Natural Language Processing
(NLP) technologies, though advanced, may struggle with interpreting complex,
ambiguous, or contextually rich queries, leading to inaccurate or irrelevant responses.

2. Data Privacy and Security: Given that users will interact with personal and
potentially sensitive documents, ensuring data privacy and security is paramount. The
system must securely handle, process, and store documents to protect against
unauthorized access, data breaches, or leaks. Building user trust requires rigorous data
protection measures and transparency about how data is used and safeguarded.

3. Handling Ambiguity and Context: Documents often contain information that is

open to interpretation or requires context for full understanding. The AI system may
find it challenging to handle queries that require subjective judgment, contextual
knowledge, or background information not contained within the document itself.
Developing mechanisms to address ambiguity and provide contextually relevant
responses is a significant challenge.

4. Website Hyperlink Parsing Limitation: When interacting with websites, the system
may encounter limitations in parsing hyperlinks within the page content. Parsing only
the specified link provided by the user may restrict the depth of content analysis and
limit the system's ability to access additional relevant information or context from
linked pages.

5. Adaptation to User Learning Styles and Preferences: Individuals have varied

learning styles and preferences for how they interact with and comprehend
information. The AI companion's ability to adapt to these differences, personalize
interactions, and provide tailored explanations is crucial for enhancing the learning
and user experience. Achieving this level of personalization involves complex user
modeling and adaptive learning algorithms.
 REFERENCES

[1] Tarun Lalwani, Shashank Bhalotia, Ashish Pal, Shreya Bisen, Vasundhara Rathod,
IEEE International Conference on Computational Systems and Information
Technology for Sustainable Solutions 2017.

[2] Emanuela Haller, Traian Rebedea, “Designing a Chat-bot that Simulates an

Historical Figure”, IEEE Conference Publications, July 2013.

[3] Amey Tiwari, Rahul Talekar, Prof. S. M. Patil, “College Information Chatbot
System”, International Journal of Engineering Research and General Science, Volume
2, Issue 2, April 2017.

[4] C. M. Bishop. Pattern recognition and machine learning. Springer New York, 2006

[5] D. M. Blei, A. Y. Ng, and M. I. Jordan. Latent Dirichlet Allocation. Journal of

Machine Learning Research, 3:993-1022, 2003.

[6] Kreimeyer K., Foster M., Pandey A., Arya N., Halford G., Jones S.F., Forshee R.,
Walderhaug M., Botsis T. Natural language processing systems for capturing and
standardizing unstructured clinical information: A systematic review. J. Biomed.
Inform. 2017;73:14–29. doi: 10.1016/j.jbi.2017.07.012. [PMC free article] [PubMed]
[CrossRef] [Google Scholar]

[7] Chowdhury G.G. Natural language processing. Annu. Rev. Inf. Sci.
Technol. 2003;37:51–89. doi: 10.1002/aris.1440370103. [CrossRef] [Google Scholar]

[8] Choudhary N (2021) LDC-IL: the Indian repository of resources for language
technology. Lang Resources & Evaluation 55:855–867. [google scholar]

[9] Bangalore S, Rambow O, Whittaker S (2000) Evaluation metrics for generation.

In proceedings of the first international conference on natural language generation-
volume 14 (pp. 1-8). Assoc Comput Linguist

[10] Baud RH, Rassinoux AM, Scherrer JR (1992) Natural language processing and
semantical representation of medical texts. Methods Inf Med 31(2):117–125

[11] Bengio Y, Ducharme R, Vincent P (2001) A neural probabilistic language model.

Proceedings of NIPS.