cpearce95-blog - Tumblr blog

[cpearce95-blog]

Project Log Entry (Entry 9)

Week 9 25.04.19 – 02.05.19:

Index

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

UX – User Experience

RNN – Recurrent Neural Network

[E] – Encoder

[D] – Decoder

This week is the final week of the project and has entailed the finalisation of the project prototype solution and deliverables prior to handover of the project deliverables in-line with the project plan. This occurred during the concluding team meeting whereby all team members were present and a lengthy discussion took place mutually evaluating the overall running of the project. Following organisation of the meeting via verbal communnicative means, the pm presented the project deliverables to the PS and AA, explaining the current state of the application in thorough detail to transfer a firm understanding of the application being transferred to ownership of the PS. Additionally, clarifying the elements of the project that are not included within the scope such as ongoing development or maintenance. Following presentation of the delierables and evaluative discussion, the PS was able to express his thoughts to the PM, discussing numerous aspects of the project such as the initiation, change control initatied, application testing and further development. Additionally, the PS confirmed his overall approval of the project meeting the requirements and outlined scope, therefore, signing the sponsor approval and sign off document, thus confirming the completion of the project.  

Week Summary

Handover of project deliverables in line with scope

Project Sponsor approval and sign off received

Final team meeting conducted

Final evaluative thoughts expressed regarding project as a whole

Meeting Minutes

Meeting Information

Objective:

To communicate the current status of the project application and handover the project deliverables to the project sponsor. Additionally, to conduct an evaluative discussion of the entire project and receive the project sponsor’s sign off and approval.

Date:

02/05/2019

Location:

Society Cafe, 5 Kingsmead Square, Bath.

Time:

16:00

Meeting Type:

Conclusive project meeting.

Called By:

Project Manager

Facilitator:

Project Manager

Note Taker:

 Administration Assistant

Attendees:

Project Manager, Project Sponsor, Administration Assistant

Agenda Items

Presenter

Time Allotted

1

Evaluative discussion of the final project prototype solution.

Project Manager

30 Minutes

2

Discussion of deliverables and elements not included.

Project Manager

10 Minutes

3

Feedback from PS received regarding the entire project.

Project Manager

20 Minutes

4

Project documentation and deliverables transferred to project sponsor.

Project Manager

25 Minutes

5

Project sponsor approval and sign off documentation completed by the project sponsor.

Project Sponsor

10 Minutes

Decisions

1

Following evaluative discussion, PS approval and sign off for the project received.

Other Notes &Information

Today's meeting has been an essential event within the project. The meeting began with a presentation of the final application by the project manager, each element of the application was presented before being the subject of evaluative team discussion. Input from the project sponsor aided in gaining full comprehension of the final prototype solution. The project manager addressed key areas of the project inclusive of project documentation for the application, and clarifying aspects not included within the scope of the project such as ongoing maintenance and development. Upon completion of the project managers presentation, the conversation moved towards possible improvements and future direction of the application. Following this, the project sponsor was presented with and signed the approval and sign off documentation, thus cementing the conclusion of the project. Following all documentation, deliverables, and software being transferred to the project sponsor, the meeting was concluded.

[cpearce95-blog]

Project Log Entry (Entry 8)

Week 8 18.04.19 – 25.04.19:

Index

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

UX – User Experience

RNN – Recurrent Neural Network

[E] – Encoder

[D] – Decoder

The events occurring throughout this week have been highly progressive towards achieving a successful project conclusion for all parties wih an interest in achieving such an income. The primary focus of the week started with the completion of hyperparameter tuning for the ChatBot application, completed by the PM as research conducted alongside aided the PM with documentative tools throughout the web. The efforts of research conducted by the PM have been recognised by the PS with a verbal form of communication taking place in the early half of the week following the completion of hyperparameter tuning, whereby the PS commmended the continuous efforts of the PM within the context of the project. The team atmosphere recognises that the project is drawing to a close and the importance of effective time-management is more prevalent than ever before within the project. The completion of application development phase has progressively transitioned into testing for the application, due to the shortage of previous experience in this specific field, the PM has researched the testing of chatbot applications, drawing reference to this project where possible and relaying this new found information into his practical actions when testing the application. Further significant project-related vents of the week have seen the penultimate team meeting take place, entailing a thorough discussion of the projects current position with an almost evaluative take on the events leading up to this point within the project lifecycle. As the project moves towards the final week, coherent understanding is held unanimously throughout the project team, the final week will entail the finalisation of the application and deliverables before the handover and project sponsor approval process takes place. Following in-depth research into chatbot testing techniques such as the pyramid approach, general test and domain specific testing, testing of the application was conducted entailing the inclusion of the following

1. Conversation Design Testing

2. Entities Testing

3. Fulfilment Testing

4. User Acceptance Testing (UAT)

Week Summary

Completion of Model Hyperparameter tuning

Commencement of application testing phase

Team meeting held whereby prototype application presented and evaluative feedback received

Meeting Minutes

Meeting Information

Objective:

To communicate the ongoing progress of the project and discuss research into and practical application of chatbot testing. Additionally, to present the prototpe application, and raise any potential concerns, complications, or issues.

Date:

22/04/2019

Location:

44 Stall St, BA1 1QH, Bath.

Time:

11:30

Meeting Type:

Penultimate team meeting; discussion and update

Called By:

Project Manager

Facilitator:

Project Manager

Note Taker:

 Administration Assistant

Attendees:

Project Manager, Project Sponsor, Administration Assistant

Agenda Items

Presenter

Time Allotted

1

Discussion of PM’s research into ChatBot testing.

Project Manager

30 Minutes

2

Discussion of testing for the project application

Project Manager

15 Minutes

3

Feedback from PS regarding the testing conducted

Project Sponsor

15 Minutes

4

Presentation of current state of project application protype prior to entering final week of project.

Project Manager

25 Minutes

5

Project discussion of next week events and arrangement of project closure meeting.

Project Sponsor

10 Minutes

Other Notes &Information

The events of this meeting predominantly centred around research conducted into chatbot application testing, before discussing the project application testing conducted. Additionally, the meeting ensured a coherent understanding of the current stage of the project by all team members as the project moves towards closure and it's final week. The project manager presented his findings from research conducted into chatbot application testing, before providing a descriptive report of the testing conducted for the project application. The project sponsor was making notes during this time and offered feedback where he felt appropriate to do so. Following discussion of testing elements including the importance of user acceptance testing, the project manager presented the prototype application to the project team in its current state, whilst discussing the finalised actions that will be taken before the project comes to a closure and deliverables are prepared and transferred.

References/Wider Reading:

Treml, F. (2018). How to Test a Chatbot - Part1: Why Is It so Hard?. Chatbots Magazine. [online] Available at: https://chatbotsmagazine.com/how-to-test-a-chatbot-part-1-why-is-it-so-hard-10f1ee8ca37d.

Al, A. (2018). How to test a Chatbot. [online] Medium. Available at: https://medium.com/@go.ako.ai/https-akoai-medium-com-how-to-test-a-chatbot-427c55365871.

AImultiple (2019). Top chatbot testing techniques and frameworks [2019 update]. aimultiple. [online] Available at: https://blog.aimultiple.com/chatbot-testing-frameworks/.

UserTesting (2019). Healthcare Chatbot Apps are on the Rise but the Overall Customer Experience (CX) Falls Short According to a UserTesting Report. [online] UserTesting. Available at: https://www.usertesting.com/about-us/press/press-releases/healthcare-chatbot-apps-on-the-rise-but-cx-falls-short.

Upwork (2019). Chatbot Testing: How to Get It Right the First Time Around. [online] upwork. Available at: https://www.upwork.com/hiring/for-clients/chatbot-testing-get-right-first-go/.

Teimao, R. (2019). How do you test Chatbots?. Quora. [online] Available at: https://www.quora.com/How-do-you-test-chatbot.

Brooke, S. (2017). How Chatbots Will Shape the Future of Healthcare. Chatbots Magazine. [online] Available at: https://chatbotsmagazine.com/how-chatbots-will-shape-the-future-of-healthcare-fa8e30cebb1c.

[cpearce95-blog]

Project Log Entry (Entry 7)

Week 7 11.04.19 – 18.04.19:

Index

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

UX – User Experience

RNN – Recurrent Neural Network

[E] – Encoder

[D] – Decoder

The events of this week have witnessed further progression and completion of the application Seq2Seq model, in addition to the planning and initiating of hyperparameter tuning for the application. Telephone calls occurred between team members to ensure a coherent understanding of the project is held. The development of the Seq2Seq model was completed following the building of the decoder model, and connection of encoder and decoder models. Research has been conducted into hyperparameter tuning within machine learning, including methods for optimising hyperparameters; grid search and random search. This research has then been placed into practical action with the project manager selecting tunable hyperparameters and object metric for the dataset. The automatic model tuning process then searches the selected hyperparameters to find the combination of values resulting in the model that optimises the objective metric. Upon completion of the hyperparameter tuning process, the developmental phase of the project and the natural language application will progress into the testing phase. The change request from the previous week has been accommodated and actioned upon. As the project progresses towards week 8, the priorities are centred around setting up and testing the application before finalization and handover of the project deliverables. Additionally, a scheduled meeting will take place between all project team members. At this stage within the project lifecycle, the PS is aware of the workload experienced by the PM throughout various other units and has provided verbal support and confirmation of their desire for the project to be seen through to completion in the following few weeks.

Week Summary

Completion of Seq2Seq Application Model

Model Hyperparameter Tuning

Completion of Development Phase

Project Communication occurring via Telephone means

References/Wider Reading

Nag, D. (2016). seq2seq: the clown car of deep learning. [online] Medium. Available at: https://medium.com/@devnag/seq2seq-the-clown-car-of-deep-learning-f88e1204dac3.

Kostadinov, S. (2019). Understanding Encoder-Decoder Sequence to Sequence Model. [online] Towards Data Science. Available at: https://towardsdatascience.com/understanding-encoder-decoder-sequence-to-sequence-model-679e04af4346.

Paul, S. (2018). Hyperparameter Optimization in Machine Learning. [online] DataCamp Community. Available at: https://www.datacamp.com/community/tutorials/parameter-optimization-machine-learning-models.

MicrosoftAzure (2018). Tune hyperparameters for your model - Azure Machine Learning service. [online] Docs.microsoft.com. Available at: https://docs.microsoft.com/en-us/azure/machine-learning/service/how-to-tune-hyperparameters.

Csaky, R. (2017). Deep Learning Based Chatbot Models. [ebook] Budapest University of Technology and Economics. Available at: https://pdfs.semanticscholar.org/f742/138e6baaecf1ee2331268917a34ebc7e6c4b.pdf.

Deshpande, A. (2019). How I Used Deep Learning to Train a Chatbot to Talk Like Me. Github.io. [online] Available at: https://adeshpande3.github.io/How-I-Used-Deep-Learning-to-Train-a-Chatbot-to-Talk-Like-Me.

[cpearce95-blog]

Project Log Entry (Entry 6)

Week 6 04.04.19 – 11.04.19:  

Index

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

UX – User Experience

RNN – Recurrent Neural Network

[E] – Encoder

[D] – Decoder

This week has witnessed further progression within the developmental phase of the ChatBot applications lifestyle. The PM has conducted in-depth research surrounding SEQ2SEQ models and recurrent neural networks (RNN’s). Having obtained a further understanding of the required steps within this phase of the project lifecycle, placeholders (variables) were created to be assigned to the data, accommodating the creation of operations and the construction of a computation graph without needing the data. Within TensorFlow terminology, this may then be fed into the graph through utilisation of the placeholders. Creation of the Seq2Seq model was comprised of two sub-models, the encoder [E] and the decoder [D]. The encoder takes in raw input text data similarly to any other RNN architecture, and outputs a neural representation. The output of the encoder model comprises the input for the decoder model, the decoder has the ability to view inside the encoders output and create differentiating output data. The steps taken to build this model are summarised in the following list.

1.       Defining input parameters for encoder model

2.       Building encoder model

3.       Defining input parameters for the decoder model

4.       Building decoder model for training

5.       Building decoder model for inference

6.       Combining decoder training/inference layer to create decoder

7.       Connecting the encoder and decoder models

8.       Defining loss function, optimiser, and application of gradient clipping

Additionally, a team meeting has been conducted this week to communicate the progress achieved thus far within the project lifecycle, in addition to raising any potential concerns held by the project manager, project sponsor, or administrative assistant. The project manager has expressed the challenges being faced throughout the development of the application solution to the project team in line with best practice communicative standards. This information that has given rise to the initiation of a change management protocol within the project in order to optimise the likelihood of a prototype solution being delivered to the project sponsor within the specified project duration.  The following section provides information concerning the team meeting held this week.  

Week Summary

·         Updated communication of project progress provided to project team via team meeting

·         Change request form submitted by Project Manager

·         Change request form subjected to approval from the Project Sponsor

·         Adjustments made to the project scope and budget to accommodate and reflect changes occurring from change management protocol

·         Building of the Seq2Seq model initiated and communicated to the project team

Meeting Minutes

Meeting Information

Objective:

To communicate the  ongoing progress of the project and update the project team to ensure  coherent understanding. Additionally, to raise any potential concerns,  complications, or issues occurring throughout project.

Date:

07/04/2019

Location:

Caffe Nero, 11 Old  Bond St,  Bath.

Time:

10:30

Meeting Type:

Team discussion and  update

Called By:

Project Manager

Facilitator:

Project Manager

Note Taker:

 Administration Assistant

Attendees:

Project  Manager, Project Sponsor, Administration Assistant

Agenda Items

Presenter

Time Allotted

1

Discussion  of PM’s research into building Seq2Seq models

Project  Manager

30  Minutes

2

Discussion  of PM’s research into existing healthcare applications and the direction of  the app.

Project  Manager

10  Minutes

3

Feedback  from PM regarding current position of project, evaluating progress and any recommendations  for more efficient development from this point.

Project  Manager

20  Minutes

4

Feedback  from PM delivered to team expressing concerns in terms of time restriction, giving  rise to a change request to make appropriate adjustments, thus, increasing  the feasibility of the project solution being produced and delivered to the project  team.

Project  Manager

25  Minutes

5

Project  sponsor approval of change request form from the project manager as it is  felt that accommodating this change will lessen certain complications with  the solution development, additionally accommodating a feasible completion of  the project.

Project  Sponsor

10  Minutes

Decisions

1

Following  discussion, Project Manager has completed a change request form for the  project, primarily to feasibly adapt to developmental delays and accommodate  the project requirement specification to the greatest extent.

2

Decision  from Project Sponsor to accommodate the change request form submitted by the  project manager, this decision was made following an in-depth evaluative discussion  of the possible effects of the change request on the overall project outcome.  

3

To  make appropriate adjustments to the project budget in order to accommodate the  change request made and alterations to the specified requirement  specification.

Other Notes  &Information

An  important meeting within the life cycle of the project as numerous elements  were subjected to discussion and key decisions made regarding the submission  and acceptance of a change request, triggering the project change management  protocol to accommodate said request. Research and findings into the  development of Seq2Seq was relayed to by the PM to the project team making  specific references to the Seq2Seq model for the project. Following this, a  general discussion took place entailing the current position of the project  as the project enters its final phases throughout the following few weeks. The  completed project change request form impacted the final project application  solution and can be found within the main report for the project. As the  project moves into week 7, the PM aims to complete production of the Seq2Seq  model before training the model with hyper parameter tuning.  

References/Wider Reading

Brijesh (2019). Seq2Seq Model using TensorFlow - knowledge Transfer. [online] Knowledge Transfer. Available at: http://androidkt.com/seq2seq-model-tensorflow.

GitHub (2017). subho406/Sequence-to-Sequence-and-Attention-from-scratch-using-Tensorflow. [online] GitHub. Available at: https://github.com/subho406/Sequence-to-Sequence-and-Attention-from-scratch-using-Tensorflow.

GitHubGist (2018). Ways to do gradients clipping and learning rate decay in tensorflow. [online] Github Gist. Available at: https://gist.github.com/raytroop/17c4b97b6a4b69d19ed164392cf39703.

Lan, H. (2019). Custom TensorFlow Loss Functions for Advanced Machine Learning. [online] Towards Data Science. Available at: https://towardsdatascience.com/custom-tensorflow-loss-functions-for-advanced-machine-learning-f13cdd1d188a.

Packt_Pub (2019). Implementing a Sequence-to-Sequence Model. [online] Hacker Noon. Available at: https://hackernoon.com/implementing-a-sequence-to-sequence-model-45a6133958ca.

Patnaik, A. (2018). Loss Function in TensorFlow. [online] Medium. Available at: https://medium.com/datadriveninvestor/loss-function-in-tensorflow-b7eb1215ef78.

Programcreek (2019). Tensorflow.clip_by_norm Python Example. [online] Programcreek.com. Available at: https://www.programcreek.com/python/example/90329/tensorflow.clip_by_norm.

StackOverflow (2019). How to apply gradient clipping in TensorFlow. [online] Available at: https://stackoverflow.com/questions/36498127/how-to-apply-gradient-clipping-in-tensorflow.

StackOverflow (2016). How to write a custom loss function in Tensorflow?. [online] Stack Overflow. Available at: https://stackoverflow.com/questions/34875944/how-to-write-a-custom-loss-function-in-tensorflow.

[cpearce95-blog]

Project Log Entry (Entry 5)

Week 5 28.03.19 – 04.04.19:  

Index:

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

UX – User Experience

RNN – Recurrent Neural Network 

The events of this week can be regarded as highly significant in their contributions towards completion of the project within the specified scope. The early stages of the week saw the completion of the UX flow design for the project application, information regarding this application was communicated to the PS via telephone means. Following discussion, verbal approval from the PS to proceed with the next steps in the developmental lifecycle process. Additionally, the week has entailed the PM obtaining a more thorough understanding and familiarity of the Linux Ubuntu OS that is now being utilised with respect to the GUI layout and terminal command use. Furthermore, the development phase of the application has been initiated with TensorFlow environment installed on the development system, datasets imported with dictionaries and lists being created. Additionally, texts have been cleaned for questions and answers before filtering the Q and A and creating dictionaries for mapping. The current stages of development have been communicated to both the PM, and AA, thus all members of the project team hold a coherent understanding of the updated project progress. Next week is going to be approached with the intent of progressing further throughout the developmental process, with a view to develop the Seq2Seq model (i.e brain of application) by creating placeholders and preprocessing targets before creating both the encoder RNN and decoder RNN for the application. 

Week Summary:

UX design flow design completion

Architecure of Chatbot discussed and subjected to confirmed PS approval

Initiation of development phase

Updated communication of project progress provided to project team

References/Wider Reading:

DataScience (2016). Tokenize sentence based on a dictionary. [online] Data Science Stack Exchange. Available at: https://datascience.stackexchange.com/questions/10424/tokenize-sentence-based-on-a-dictionary.

Gk (2017). Contextual Chatbots with Tensorflow. Chatbots Magazine. [online] Available at: https://chatbotsmagazine.com/contextual-chat-bots-with-tensorflow-4391749d0077?gi=f0d5ef2beb22.

StackOverflow (2015). Create a dictionary for the token, tags of a text. [online] Stack Overflow. Available at: https://stackoverflow.com/questions/27177165/create-a-dictionary-for-the-token-tags-of-a-text.

TensorFlow (2019). Install TensorFlow with pip  |  TensorFlow. [online] TensorFlow. Available at: https://www.tensorflow.org/install/pip.

TensorFlow (2019). Adding a dataset  |  TensorFlow Datasets  |  TensorFlow. [online] TensorFlow. Available at: https://www.tensorflow.org/datasets/add_dataset

Zuppichini, F. (2018). How to use Dataset in TensorFlow. [Blog] Towards Data Science. Available at: https://towardsdatascience.com/how-to-use-dataset-in-tensorflow-c758ef9e4428.

Tensorflow (2019). Datasets for Estimators  |  TensorFlow Core  |  TensorFlow. [online] TensorFlow. Available at: https://www.tensorflow.org/guide/datasets_for_estimators.

[cpearce95-blog]

Project Log Entry (Entry 4)

Week 4 21.03.19 – 28.03.19:

Index:

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

UX – User Experience

OS - Operating System

The activities of this week in relation to project predominately centred around the design of the user experience (UX) entailing more in depth research to assist in the design process. Research into chatbot design highlighted strengths in various modern applications that equate to a user flow that is logical and efficient for the user, additionally, research outlined applications which have provided a lack of clarity of users and making the achievement of the desired outcome more challenging. Elements of this research have been considered and where relatable, incorporated into the initial user flow of the application for this project to optimise usability and the logical flow. Further to this, the project manager has changed operating system running on his home desktop due to problems occuring with Windows 10, the application will now be developed on a Linux Distribution Ubuntu 18.0.4 OS system. The problems were causing the computer to perform very slowly which if allowed to continue, would impact the time control of the project. This information was communicated to the PS and AA via telephone to ensure coherent understanding. An additional activity of the week entailed a project team meeting whereby the week events and project progress was communicated and discussed, details of which can be found in the meeting information section below. 

Week Summary:

⦁ Uninstall of Windows 10 OS and Installation of Linux Distribution Ubuntu v-18.0.4 for development, triggering possible delay and need for change management protocol activation

⦁ Further research into designing effecive user experience (UX) for Chatbot purposes

⦁ Initial User Experience (UX) of Chatbot design completion

⦁ Project team meeting conducted

Meeting Information:

Objective: To communicate the ongoing progress of the project and update the project team to ensure coherent understanding. Additionally, to raise any potential concerns, complications, or issues occuring throughout project. 

Date: 24/03/2019 Location: Boston Tea Party, 8 Alfred St, Bath. 

Time: 14:30  Meeting Type: Team discussion and update

Called By: Project Manager Facilitator: Project Manager

Note Taker:   Administration Assistant

Attendees: Project Manager, Project Sponsor, Administration Assistant

Agenda Items Presenter Time Allotted

1 Discussion of PM’s research into User Experience and Flow Design of Chatbots.  Project Manager 30 Minutes 

2 Windows 10 OS uninstall and Linux Ubuntu System install communicated to team.  Project Manager 10 Minutes

3 Feedback from PM regarding early stages of Chatbot UX flow design.  Project Manager 20 Minutes

4 Example conversational scripts discussed and evaluated by project team.  Project Manager 25 Minutes

5 Project sponsor approval of current level of research conducted and teams commitment to achieving project aims and objectives. Project Sponsor 10 Minutes

Decisions

1 Following discussion, PM approval of UX design format received encouraging further progress of project application design and development. 

2 Decision to consider possible reduction in requirement specification due to technological complications with PM's home system requiring new OS install, adjustment would increase overall probability of project completion matching the objectivies within specified time frame. 

3 To optimise functionality, usability and logical flow for user experience opposed to spending excessive time resources on GUI of application. 

Other Notes & Information

A productive meeting with some in depth discussions taking place regarding UX design of application, the PM brought current industry applications

to the attention of the team highlighting positive aspects and elements critiqued where applicable. The general consensus throughout the project team was in approval of the draft user flow presented to the PS by the PM. The UX draft depicted a system whereby a user would sign up at the request of the chabot, some basic information regarding the individual such as name, gender, age, and certain health habits would be obtained (e.g whether smoker or known to have high blood pressure). Following this the application design flow provided a sympton assessment process whereby a report will be formed relative to the information provided by the user, this report would then deliver a summary section with reccommended advice in terms of receiving a diagnosis or prevention or further symptons. The conceptual structure of the application at this stage is firmly understood and approved by the entirety of the project team. Additionally, the PM has expressed possible concern about achieving the entire requirement specification within the specified time scale for the project due to various factors with the most identifiable being a) Current lack of experience in the field, b) Uninstall of Windows 10 on development System due to problems, installing alternative Ubuntu OS for development, and c) Current work commitments to other tasks and employment. These concerns have been communicated to the team giving rise to a possible change management protocol, however, at this stage the project remains on track for completion, thus, these concerns were communicated to mitigate the risks of breaching the time scale of the project or not achieving full requirements within project outcome. This aspect of the project will continue to be monitored by the PM and communicated to the PS as the project moves forwards towards development. 

References/Wider Reading:

Fadhil, A. (2018). A Conversational Interface to Improve Medication Adherence: Towards AI Support in Patient’s Treatment. [ebook] Trento: University of Trento. Available at: https://arxiv.org/ftp/arxiv/papers/1803/1803.09844.pdf.

Duijst, D. (2017). Can we Improve the User Experience of Chatbots with Personalisation?. [ebook] Amsterdam: University of Amsterdam. Available at: https://www.researchgate.net/publication/318404775_Can_we_Improve_the_User_Experience_of_Chatbots_with_Personalisation.

Fadhil, A. and Schiavo, G. (2019). Designing for Health Chatbots. [ebook] Trento: University of Trento. Available at: https://www.researchgate.net/publication/331343433_Designing_for_Health_Chatbots.

Smestad, T. (2018). Personality Matters! Improving The User Experience of Chatbot Interfaces. [ebook] Norwegian University of Science and Technology. Available at: https://brage.bibsys.no/xmlui/bitstream/handle/11250/2502575/18507_FULLTEXT.pdf?sequence=1&isAllowed=y.

Valtolina, S., Barricelli, B., Di Gaetano, S. and Diliberto, P. (2018). Chatbots and Conversational Interfaces: Three Domains of Use. [ebook] Milan: University of Milan - Dept of Computer Science. Available at: http://ceur-ws.org/Vol-2101/paper8.pdf.

Jain, M., Kumar, P., Kota, R. and Patel, S. (2018). Evaluating and Informing the Design of Chatbots. [ebook] Seattle: University of Washington. Available at: https://ubicomplab.cs.washington.edu/pdfs/chatbot_study.pdf.

[cpearce95-blog]

Project Log Entry (Entry 3)

Index:

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

Week 3 14.03.19 – 21.03.19:

This week has been another productive week for the project team and working towards a successful delivery of the project outcomes. Leading on from last weeks tasks, the PM placed time resources into further research and consideration of machine learning technologies inclusive of DialogFlow and TensorFlow platforms. Whilst the PM outlined the software to utilise as TensorFlow within the initial approved action plan, it was felt that both platforms provide useful tools to design and development the AI Chatbot, thus, further comparative research would highlight the strengths and weaknesses of each in more detail before making the most informed decision on which platform to utilise. This was communicated with the PS and received with a mutual understanding as the importance of selecting the most appropriate platform for development was emphasised by the PM. Further research was conducted into chatbot design and development using TensorFlow and DialogFlow, considering the stages of designing an effective model that correlate as close to the project outcomes as possible. Conducted research looked into Recurrent Neural Network Architectures (RNN’s) and the technological concepts within this such as the make-up of a recurrent neuron, unfolding an RNN, and the anatomy of a deep neural network. Additionally, the activities of the week saw further research conducted into use of Seq2Seq models in TensorFlow as the PM obtained a clearer understanding of the steps involved in developing a Seq2Seq model within tensorflow and communicated his findings to the team via telephone. Furthermore, a skeletal flow structure for the application has undergone creation by the PM which signifies the user experience (UX) and flow of the application from a users’ perspective. This is being performed following an analysis and evaluation of data findings accumulated during data collection activities.

Week Summary:

•                     Additional consideration and comparison of AI and ML software, Google Dialogflow and TensorFlow performed

•                     Research conducted into Recurrent Neural Network Architectures and TensorFlow Seq2Seq Models

•                     Results of data collection evaluated to mould preferences and potential design features of chatbot application

•                     Application user experience (UX) flow design started and application machine learning based intent and entity extraction investigated

References/Wider Reading:

Chansung, P. (2018). Seq2Seq model in TensorFlow. [online] Towards Data Science. Available at: https://towardsdatascience.com/seq2seq-model-in-tensorflow-ec0c557e560f.

Zhang, H., Lan, Y., Guo, J., Xu, J. and Cheng, X. (2018). Tailored Sequence to Sequence Models to Different Conversation Scenarios. [ebook] Beijing, China: Institute of Computing Technology. Available at: https://aclweb.org/anthology/P18-1137.

Linkai, L. (2017). Developing Seq2Seq Model with Tensorflow. [ebook] Deep Learning Research & Application Center. Available at:https://dlc.hsu.edu.hk/wp-content/uploads/2017/05/week11-developing_seq2seq_with_tensorflow.pdf.

Kuchaiev, O., Ginsburg, B., Gitman, I., Lavrukhin, V., Li, J., Nguyen, H., Case, C. and Micikevicius, P. (2018). Mixed-Precision Training for NLP and Speech Recognition with OpenSeq2Seq. [ebook] Santa Clara: NVIDIA. Available at: https://arxiv.org/pdf/1805.10387.pdf.

FinancesOnline (2019). Compare TensorFlow vs Dialogflow 2019 | FinancesOnline. [online] Financesonline.com. Available at: https://comparisons.financesonline.com/tensorflow-vs-dialogflow.

Couto, J. (2017). Building a Chatbot: Analysis and Limitations of Modern Platforms. [ebook] DZone. Available at: https://dzone.com/articles/building-a-chatbot-analysis-amp-limitations-of-mod.

Narwekar, A. and Pampari, A. (2016). Recurrent Neural Network Architectures. [ebook] Illinois: University of Illinois. Available at:http://slazebni.cs.illinois.edu/spring17/lec20_rnn.pdf.

Bapat, R. (2017). Helping Chatbots To Better Understand User Requests Efficiently Using Human Computation. [ebook] Delft: Delft University of Technology. Available at: http://htttp://wis.ewi.tudelft.nl.

Aspect Software (2017). 10 Steps to Chatbot Creation. [ebook] Aspect Software. Available at: https://www.aspect.com/globalassets/microsite/nlu-lab/images/10-Steps-to-Chatbot-Creation.pdf.

[cpearce95-blog]

Project Log (Entry 2)

Index:

PM – Project Manager

PS – Project Sponsor

AA – Administrative Assistant

AI – Artificial Intelligence

ML – Machine Learning

NLP – Natural Language Processing

NLTK – Natural Language Toolkit

Week 2 07.03.19 – 14.03.19:  This week has primarily focused around further research into chatbot AI design and development, additionally, the weeks’ primary tasks entailed the design of data collection instruments and completion of data collection from the sample group. The sample group was formed by utilising a random figure of males and females in addition to their age ranges, thus representing high integrity in the research. The first official team meeting was conducted whereby the findings of the project managers research in addition to the study results were communicated to the project sponsor and team. The project sponsors constructive comments were recorded in the meeting minutes document. Additionally, the developmental methodology was discussed and it was confirmed that initially proposed agile methodology will be followed throughout. Furthermore, the week consisted of research into Python programming language, specifically in reference to using the natural language toolkit (NLTK) with Python 3, one of the leading platforms for working with human language data and Python. Complex topics within this were discussed such as linear regression, classifier, creating, training and evaluating a neural network like CNN, RNN, auto encoders etc. As part of successfully delivering the project outcomes, it is imperative that the project manager optimises their overall understanding of using the NLTK platform. 

Week Summary: 

- Continued data collection instruments designed - Further research conducted into AI chatbot design and development with Python 3 - Data collection performed utilising designed collection instruments - Research into NLTK with Python

Meeting Information:

Objective: To discuss findings from research conducted by the project manager, to communicate findings from data collection conducted thus far, discuss methodology behind project developmental lifecycle

Date: 11/03/2019 Location: Caffe Nero, Dorchester St, Bath. Time: 11:00am Meeting Type: Team discussion Called By: Project Manager Facilitator: Project Manager Note Taker:  Administration Assistant Attendees: Project Manager, Project Sponsor, Administration Assistant Agenda Items Presenter Time Allotted 1. Discussion of PM’s research into AI chatbot development with Python 3 Project Manager  - 30 Minutes 2. Discussion of developmental methodology with Agile being confirmed as most appropriate Project Manager - 30 Minutes 3. Discussion of research into AI chat bot design and development with NLTK & Python 3 Project Manager - 30 Minutes 4. Discussion of Date Collected from Data Collection Instruments thus far. Project Manager - 25 Minutes 5. Project sponsor approval of current level of research conducted and teams commitment to achieving project aims and objectives. Project Sponsor - 10 Minutes

Decisions:

   1. Decision made by project team to utilize agile methodology throughout the development of the Chatbot 2. To invest significant amount of time (hours) into a full comprehension of using Python 3 programming language, particularly in respect to machine learning, natural language processing, and NLTK. 3. Satisfaction expressed from PS with the current data collection methods and findings.

Other Notes & Information:

Another good meeting for the project team, once more, the moral between the project team members appears to be fully coherent with a shared understanding of the project tasks and the methodology behind achieving overall project success. The planned agile methodology set out in the proposal following discussion, has been confirmed as the most suitable approach as the design of the application nears. The complexity of utilizing the NLTK with Python 3 has been made apparent to the project team by the project manager, however, as further research is conducted into utilizing the platform with Python 3 programming language, it is evident that the project managers understanding of the developmental lifecycle of a Chatbot incorporating natural language processing technologies is increasing, thus, granting confidence for the project sponsor.

References/Wider Reading: Stanford University (2017). A TensorFlow Chatbot. [ebook] Stanford: Stanford University. Available at: https://web.stanford.edu/class/cs20si/lectures/slides_13.pdf. Rahman, A., Mamun, A. and Islam, A. (2017). Programming challenges of chatbot: Current and future prospective. [ebook] Dhaka: IEEE Humanitatian Technology Conference. Available at: https://www.researchgate.net/publication/323211844_Programming_challenges_of_chatbot_Current_and_future_prospective. Harrison, K. (2018). Python Programming Tutorials. [online] Pythonprogramming.net. Available at: https://pythonprogramming.net/chatbot-deep-learning-python-tensorflow/. Sumi, R. (2019). Building Chatbots with Python. [ebook] Apress. Available at: https://www.apress.com/gb/book/9781484240953. Krishnan, S. (2018). Chatbots are cool! A framework using Python. [online] Towards Data Science. Available at: https://towardsdatascience.com/chatbots-are-cool-a-framework-using-python-part-1-overview-7c69af7a7439. Bird, S., Klien, E. and Loper, E. (2016). Natural Language Processing with Python. [ebook] O'reilly. Available at: http://www.datascienceassn.org/sites/default/files/Natural%20Language%20Processing%20with%20Python.pdf. Madnani, N. (2007). Getting Started on Natural Language Processing with Python. [ebook] The University of Maryland. Available at: http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.110.2178&rep=rep1&type=pdf. Cioroianu, I. (2013). Natural Language Processing in Python using NLTK. [ebook] New York University. Available at: http://www.nyu.edu/projects/politicsdatalab/localdata/workshops/NLTK_Presentation.pdf.

[cpearce95-blog]

Project Log (Entry 1)

Index: PM – Project Manager PS – Project Sponsor AA – Administrative Assistant AI – Artificial Intelligence ML – Machine Learning NLP – Natural Language Processing NLTK – Natural Language Toolkit

Week 1 28.02.19 – 07.03.19:  

The week has seen the official initiation of the project, this entailed the presentation of the project plan and proposal to the project sponsor. The project sponsor subjected the project to approval and has signed the project sponsorship approval letter outlining the expectations of the project, in addition to the regulations which must be adhered to throughout the undertaking of the project. The first team meeting was held with all members present, the expected contributions from team members were discussed in line with the project plan. Additionally, the week entailed research being conducted into AI chat bot design and development to gather a firmer understanding of the process before carrying it out. The findings of the research were evaluated and discussed with PS to formulate future direction for the application. Research was conducted into the design of effective data collection instruments to support findings from research and maintain coherent progress. The sample group from which the data will be collected was formed with the administrative assistant providing help in the form of e-mail communications with study sample members. The sample group contained a wide range of diversity with candidates completing a simple introductory questionnaire to inform the researcher of their age, gender, sexual orientation and other characteristics. The diversity throughout the sample group aims to ensure the highest levels of validity and integrity are maintained throughout the project. 

Week Summary:

- Project Initiation completed (1st Milestone) - Project plan presented to project team and subjected to approval - Research conducted into AI Chatbot design and development - Research conducted into Data collection Instruments design - Study sample group selected and contacted - Team meeting conducted with all project team present

Meeting Information:

Objective: To present project plan and proposal to project team, subject to approval initiate project. Date: 28/02/2019 Location: Costa Coffee Shop, Milsom Street, Bath. Time: 14:00 Meeting Type: Introductory/Initiation Called By: Project Manager            Facilitator: Project Manager Note Taker:  Administration Assistant Attendees: Project Manager, Project Sponsor, Administration Assistant Agenda Items Presenter Time Allotted 1. Presentation and run through of Project Proposal/Plan Project Manager -  30 Minutes 2. Discussion of project plan N/A -  30 Minutes 3. Discussion of research into AI chat bot design & development Project Manager - 15 Minutes 4. Discussion of Date Collection Instrument design Project Manager  - 15 Minutes 5. Project sponsor approval Project Sponsor -  5 Minutes

Decisions: 1. Decision made by project sponsor to continue with funding as sponsorship approval letter was signed. 2. To design data collection instruments with a diverse audience in mind, asking open questions where feasible to obtain integral qualitative information. 3. To invest time resources into obtaining further understanding of Chat Bot design and development life cycle via research.

Other Notes & Information:

Overall successful meeting with generally good moral present throughout the team. The end concept has been discussed in addition to the process of achieving the project outcome. Coherent awareness of the project objectives and tasks has been obtained by all members of the team. Project sponsor understands the potential constraints and risks in achieving project success, in addition to the steps being taken to minimize these. Moving into week 2, the project looks to conduct research via the data collection instruments, currently being designed by the project manager to obtain the most honest reflection of views from study sample group.

References/Wider Reading:

Vogel, J. (2017). Chatbots: Development and Applications. [ebook] Berlin: HTW Berlin, University of Applied Sciences. Available at: https://jorin.me/chatbots.pdf. Cahn, J. (2017). CHATBOT: Architecture, Design, & Development. [ebook] Pennsylvania: University of Pennsylvania - Department of Computer and Information Science. Available at: https://static1.squarespace.com/static/569293741c1210fdda37b429/t/59160b6bff7c50104e601a85/1494616940469/CHATBOT_thesis_final.pdf. Schlesinger, A., O'Hara, K. and Taylor, A. (2018). Let's Talk About Race: Identity, Chatbots, and AI. [ebook] Squarespace. Available at: https://static1.squarespace.com/static/5a8b405a18b27d5478196dca/t/5a8b690d24a694d7072d25a1/1519085853799/chi18-schlesinger-LetsTalkAboutRace.pdf. Raman, A. and Tok, W. (2018). A Developer's Guide to Building AI Applications: Create Your First Intelligent Bot with Microsoft AI. 1st ed. [ebook] O'Reilly Media, Inc.: Sebastopol. Available at: https://info.microsoft.com/rs/157-GQE-382/images/EN-US-CNTNT-eBook-AI-A-Developer%27s-Guide-to-Building-AI-Applications.pdf?mkt_tok=eyJpIjoiWW1ReE9HWXpaamN5TWpNMiIsInQiOiJRRWhSTjJseTJlTG5IemM5VG1yb3ZCdlY4WjVXN1MwS1FTU1NpRmxqSkNVVklJS1NYTHVTakZYeDhjWTA3dnZ1MlJyTHVpVHUxcWNRc2xHVVdycnFnV2IzXC82aVVockR3dWxFVUV6TFpUR09lV0pINGtOXC9FQnFqQnZPNG9lckRwZXVhajIzTUhMYk5QSk1hQ3dPNmRHZz09In0%3D. Ahmad, N., Hamid, M., Zainal, A., Rauf, M. and Adnan, Z. (2018). Review of Chatbotss Design Techniquesw. [ebook] International Journal of Computer Applications. Available at: https://www.researchgate.net/publication/327097910_Review_of_Chatbots_Design_Techniques. SAP (2018). The Future of Chatbots Report: How Chatbots Can Improve the Customer Journey. [ebook] SAP Hybris. Available at: https://cx.sap.com/medias/sys_master/root/hc1/hdb/8828833824798/whitepaper-the-future-of-chatbots-en.pdf?campaigncode=undefined.