The session will identify what applications and opportunities AI and ML present and how anyone can get started, there is a whole host of free resources to start learning and experimenting. One of the repetitive tasks which can be automated is the categorisation of keywords which can be sped up using supervised models. Not only is it fascinating to understand what is capable, but the applications mean that you can free your time up to spend more time on tasks which adds more value for your clients/business.

1. Leveraging AI/ML to
Automate the
Repetitive Tasks
Sabrina Bandel | The7Stars
SLIDESHARE.NET/SabrinaBandel
@sabrina_bandel

2. Agenda
• Machine Learning Examples
• Automating Keyword
Categorisation
• Tools & Resources

3. Improvements in
technology have led to
advancements in
automation

4. Free Resources To Get Started – Classify
Images of Clothing
Fashion MNIST dataset which contains
70,000 grayscale images in 10 categories
https://colab.research.google.com/github/tensorflow/docs/blob/master/site/en/tutorials/keras/classification.ipynb#scrollTo=yR0EdgrLCaWR
Label Class
0 T-shirt/top
1 Trouser
2 Pullover
3 Dress
4 Coat
5 Sandal
6 Shirt
7 Sneaker
8 Bag
9 Ankle boot
#brightonSEO

5. Plotting Predictions to Verify Results
Label Class
0 T-shirt/top
1 Trouser
2 Pullover
3 Dress
4 Coat
5 Sandal
6 Shirt
7 Sneaker
8 Bag
9 Ankle boot
#brightonSEO

6. Results Can Be Incorrect Even When
Confidence is High
Label Class
0 T-shirt/top
1 Trouser
2 Pullover
3 Dress
4 Coat
5 Sandal
6 Shirt
7 Sneaker
8 Bag
9 Ankle boot
#brightonSEO

7. Reducing the Time it Takes to Write Meta
Descriptions for Large Websites
Use Text Summarization Algorithms to Help Aid
the Writing of Meta Descriptions (GitHub Repo)
#brightonSEO

8. What Are the Repetitive Tasks That You
Undertake?
• Forecasting
• Classification
• Meta data
• Image alt text
• Sentiment analysis
#brightonSEO

9. Automating
Keyword
Categorisation

10. The aim of keyword
research is to understand
your audience’s needs and
behaviour.

11. Keyword Research Underpins Many Tasks
• Identifying content gaps
• Optimising and refreshing content
• Keyword mapping and strategy
• Gain audience insight
• Organise content in a user-friendly information architecture
• Segment client performance in search by category for keyword tracking
#brightonSEO

12. When You Need to Categorise Thousands
of Keywords
Phones
iPhone Samsung Google
iPhone 12 Pro
Max
iPhone 12 Pro
iPhone 12
iPhone 12 Mini
Galaxy Z Flip
Galaxy Z Fold
Galaxy S21
Galaxy S20
Pixel 5a
Pixel 5
Pixel 4a
Pixel 4
#brightonSEO

13. Keyword Categorisation is a Time Intensive
Process
Insight &
Analysis
Categorisation
Cleaning &
Sorting
Keyword List
A list of a few thousand keywords can take a few days to categorise,
depending on the level granularity.
#brightonSEO

14. How do we
speed up the
process?

15. Apply Machine Learning Techniques to
Automate the Process
Artificial Intelligence
Machine Learning
Deep Learning
Artificial Intelligence
A technique which enables
machines to mimic human
behaviour
#brightonSEO

16. Apply Machine Learning Techniques to
Automate the Process
Artificial Intelligence
Machine Learning
Deep Learning
Machine Learning
Subset of AI which uses statistical
methods to enable improvements
with experience
#brightonSEO

17. Apply Machine Learning Techniques to
Automate the Process
Artificial Intelligence
Machine Learning
Deep Learning
Deep Learning
Subset of ML which make
computation of multi-layer neural
network feasible
#brightonSEO

18. Supervised or Unsupervised Learning?
Machine Learning
Supervised Learning Unsupervised Learning
Regression Classification Clustering
Develop predictive models
based on both input and
output data
Group and interpret data
based only on input data
#brightonSEO

19. How Does Unsupervised Machine Learning
Work?
Input Data Prediction
#brightonSEO

20. How Does Supervised Machine Learning Work?
Input Data
Annotations
These are t-
shirts
Prediction
It is a t-shirt
#brightonSEO

21. What is the Process?
3,000 Keywords
Keywords to Label
750 Keywords
Keywords to Predict
2,250 Keywords
#brightonSEO

22. Cleaning Natural Language Data
Tokenization
Tokenization is splitting a phrase, sentence,
paragraph, or an entire text document into
smaller units, such as individual words or terms.
E.g. ‘natural language processing’
becomes ‘natural’ ‘language’ ‘processing’
3,000 Keywords
Keywords to Label
750 Keywords
Keywords to Predict
2,250 Keywords
#brightonSEO

23. Cleaning Natural Language Data
Tokenization
3,000 Keywords
Keywords to Label
750 Keywords
Keywords to Predict
2,250 Keywords
Lemmatization
Lemmatization reduces a given word to its root
word. The root word is called a lemma in the
lemmatization process
E.g. ‘insurance’ and ‘insure’ become ‘insure’
#brightonSEO

24. How Does a Machine Process Natural
Language?
clothing
shirt
jeans
dress
trainers
jeans
shirt jeans dress trainers
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
0 1 0 0
A machine cannot read categories only numerical data. Therefore all input data must be
converted into numerical data, a common method is to use one-hot encoding.
#brightonSEO

25. Cleaning Natural Language Data
Tokenization &
Lemmatization
3,000 Keywords
Keywords to Label
750 Keywords
Keywords to Predict
2,250 Keywords
Training Set
80%
Test Set
20%

26. Apply the Supervised Model
Input Data
Annotations
These are t-
shirts
Prediction
It is a t-shirt
Model
Training
Processing
Test Data
#brightonSEO

27. Model Performance Metrics
• Accuracy – Out of all predictions
made, how many are correctly
classified?
• Precision - What proportion of
positive identifications were actually
correct?
• Recall (sensitivity) - What proportion
of actual positives was identified
correctly?
#brightonSEO

28. How to Get
Started

29. What Tools Can Be Used to Automate Tasks?
#brightonSEO

30. Online Forums Help to Solve Errors
#brightonSEO

31. Key Takeaways
• AI and Machine Learning techniques can speed up processes.
• If you have a time intensive and repetitive task it can probably be
automated.
• Open source resources means you do not need to have a deep
knowledge.
• Anyone can learn R or Python. When you encounter a problem paste the
error message in Google.
#brightonSEO

32. Thank You
Sabrina Bandel
Senior SEO Executive
@sabrina_bandel