You're dealing with shrinking budgets, disappearing clients, and taking on the work of furloughed coworkers. How do you continue to deliver amazing results with limited time and resources?
Writing quality content that educates and persuades is still a surefire way to achieve your traffic and conversion goals. But the process is an arduous, manual job that doesn't scale.
Fortunately, the latest advances in Natural Language Understand and Generation offer some promising and exciting results.
Hamlet will walk you through what is possible right now using practical examples (and code!) that technical SEOs can follow and adapt for their business.
4. You are Leveraging AI in Your
SEO Work Already
Using Google
Docs or Gmail
Smart
Compose?
@hamletbatista#SEJLive
5. Facing Writers’ block?
Start typing
and hit the
tab key for
full sentence
ideas!
https://transformer.huggin
gface.co/doc/distil-gpt2
@hamletbatista#SEJLive
6. How to Go Deeper with Keyword Research
“We are in the era where intent-based searches are more
important to us than pure volume.”
“You should take the extra step to learn the questions
customers are asking and how they describe their problems.”
“Go from keywords to questions”
-Mindy Weinstein
SEJ / May 9, 2020
https://www.searchenginejournal.com/deep-keyword-research-process/214096/
@hamletbatista#SEJLive
7. What is the Opportunity?
1. Search engines are
answering engines these
days.
2. One effective way to write
original, and popular
content is to answer your
target audience’s most
important questions.
@hamletbatista#SEJLive
8. 3. FAQ search snippets
take more real estate in
the SERPs.
4. Doing this manually is
going to be expensive
and time consuming.
@hamletbatista#SEJLive
What is the Opportunity?
9. 5. Let’s automate it by
leveraging AI and existing
content assets
@hamletbatista#SEJLive
What is the Opportunity?
10. Leveraging Existing Knowledge
1. Most established
businesses have
valuable knowledge
bases
2. Many times not yet
publicly available
(support emails,
chats, internal wikis).
@hamletbatista#SEJLive
11. Open Source AI + Proprietary Knowledge
Using a technique called
Transfer Learning, we
can produce original,
quality content by
combining proprietary
knowledge bases and
public deep learning
models and datasets.
@hamletbatista#SEJLive
12. AGENDA
We are going to review automated question and answer generation approaches:
1. We will source popular questions using online tools
2. We will answer them using two NLG approaches:
a. A span search approach
b. A “closed book” approach
3. We will add FAQ schema and validate using the SDTT
4. Resources to learn more
@hamletbatista#SEJLive
23. In Just 3 Lines of Python Code
!pip install transformers
from transformers import pipeline
# Allocate a pipeline for question-answering
nlp = pipeline('question-answering')
nlp({
'question': 'What is the name of the repository ?',
'context': 'Pipeline have been included in the huggingface/transformers repository'
})
{'answer': 'huggingface/transformers',
'end': 59,
'score': 0.5135626548884602,
'start': 35}
@hamletbatista#SEJLive
24. A SPAN SEARCH APPROACH
● Load the Transformers NLP library
https://github.com/huggingface/transformers
● Allocate a Question Answering pipeline
https://huggingface.com/transformers/usage.html#extractive-
question-answering
● Provide the question and context (content/text most likely to
include the answer)
@hamletbatista#SEJLive
25. How to Get the Context
!pip install requests-html
from requests_html import HTMLSession
session = HTMLSession()
url = "https://www.searchenginejournal.com/uncover-powerful-data-stories-phyton/328471/"
selector = "#post-328471 > div:nth-child(2) > div > div > div.sej-article-content.gototop-pos"
with session.get(url) as r:
post = r.html.find(selector, first=True)
text = post.text
@hamletbatista#SEJLive
27. Exploring the Limits of NLG with T5
and Turing-NLG
Google’s T5 (11-billion
parameter model) and
Microsoft’s TuringNG
(17-billion parameter
model) are able answer
questions without
providing any context!
🤯🤯🤯
@hamletbatista#SEJLive
28. Open Book vs Closed Book Question Answering
@hamletbatista#SEJLive
29. Closed Book Trivia Challenge with T5
The Google’s T5 team
went head-to-head
with the 11-billion
parameter model in a
pub trivia challenge
and lost! 😅
https://t5-
trivia.glitch.me/
@hamletbatista#SEJLive
31. Here is T5AnsweringArbitrary Questions
We are going to
train the 3-billion
parameter model
using a free
Google Colab
TPU.
These are some
example
predictions.
32. ● Copy the example Colab notebook to your Google Drive
● Change the runtime environment to Cloud TPU
● Create a Google Cloud Storage bucket (use the free $300 in credits)
● Provide the bucket path to the notebook
● Select the 3-billion parameters model
● Run the remaining cells up to the prediction step
We won’t need to write any Python code 😞
HERE IS THE TECHNICAL PLAN
@hamletbatista#SEJLive
33. Copy the Colab Notebook to Your Google Drive
@hamletbatista#SEJLive
39. Run the Remaining Cells up to the Prediction Step
@hamletbatista#SEJLive
40. FINE TUNING TO ADD
PROPRIETARY
KNOWLEDGE
@hamletbatista#SEJLive
41. Add New Proprietary Training Datasets
1. Preprocess your
proprietary
knowledge base
into a format that
can work with T5
2. Adapt the existing
code for this
purpose (Natural
Questions,
TriviaQA)
42. 1. Extract
2. Transform
3. Load
https://www.searchenginejournal.com/machine-learning-practical-introduction-seo-professionals/366304/
Add New Proprietary Training Datasets
48. 1. Introduction to Python
for SEOs
2. Introduction to Machine
Learning for SEOs
3. Leverage SOTA models
with one line of code
4. Exploring Transfer
Learning with T5
5. Deep Learning on
Steroids with the Power
of Knowledge Transfer!
6. MarketMuse First Draft
50. About RankSense
Automate tedious SEO tasks in
Google Sheets.
Import the sheets and deploy
them as experiments to
Cloudflare.
Learn which changes are
effective.
https://www.ranksense.com
@RankSense#SEJLive