3. What is a trending topic?
ComInSySM.R.Khadivi - Trending topics in Social Networks 3
Definition: A trending topic is a subject that experiences a surge in
popularity on one or more social media platforms for a limited duration
of time.
4. What is a trending topic? – cont.
ComInSySM.R.Khadivi - Trending topics in Social Networks 4
Trending topic is a term coined by Twitter to refer to the most used
keywords on the social network during a given period of time. It is a
concept related to fashion trends and topics, what everyone is talking
about at any given time.
5. What is a trending topic? – cont.
ComInSySM.R.Khadivi - Trending topics in Social Networks 5
• On the social networking, the trending topics are a listing of the top
keywords or hashtags being discussed currently in a given time.
• These trends are a powerful and easy way to see what the hot topics
around the world or in your area are and who is talking about them.
• These topics can really be about anything, depending on what is
relevant at the time.
• When you join the conversation on a trending topic, you expose your
post to everyone talking about and looking at that topic.
6. What is a trending topic? – cont.
ComInSySM.R.Khadivi - Trending topics in Social Networks 6
• Trending topics often evolve around popular cultural occurrences
• There's no limit to how long a topic stays popular
• Tend to have a shelf-life of one day to one week
7. How to identify trending topics?
ComInSySM.R.Khadivi - Trending topics in Social Networks 7
• Twitter (Social media) has an algorithm
• Trending topics often come in the form of hashtags
8. Hashtags
M.R.Khadivi - Trending topics in Social Networks
• very popular tagging method
• Used to mark keywords or topics of a post
• especially visible in microblogging posts like in Twitter, Facebook, etc.
• a single word that start with ‘‘#’’ symbol
• contains only alphanumeric characters
• might be a concatenation of several words it does not contain spaces.
• no limitation which hashtag form is correct and which not
• in social networks there are hashtags that appear more often than others
ComInSyS 8
9. Trending topics
M.R.Khadivi - Trending topics in Social Networks
• The topics which are tagged with those very popular tags
• Twitter users who use those popular tags are interested in those topics
• like to have an influence on global , have to tag posts without any misspelling
• tags related to trending topics, have to be more frequent than other hashtags
• The frequency refers to:
• a frequency of hashtag in post of a single user
• frequency of hashtags in posts of different users.
• the statistical filtration model to find those trending topics among a given society
ComInSyS 9
10. Hashtags filtration model
M.R.Khadivi - Trending topics in Social Networks
list of indices of followers that have
hashtag h among their hashtags
unordered list of hashtags
that belongs to influencer i
number of hashtags of type h in unordered
list that belongs to influencer i
number of all hashtags in unordered
lists that belongs to influencer i
number of users followers that have used hashtag
h divided by the number of all followers the length (in characters) of hashtag h, excluding #
threshold values of the model
ComInSyS 10
11. Hashtags filtration model - Cont.
M.R.Khadivi - Clustering of trending topics in microblogging posts: A graph-based 11 / 25
• The role of the first threshold value is to remain only those hashtags that are present in more
than T1 percent of all posts of that person.
• The second threshold eliminates users that do not use enough hashtags to use them for
statistical computation. They have to post more than T2 hashtags.
• The third threshold governs if a hashtag is present in posts of more than T3 percentage of
distinct users. That is very important for model because we want to find not only tags that are
often used by a single or several person but in a larger group.
• The last threshold checks if length of tagging word is greater than T 4. Many of spamming or
irrelevant words have less than five letters and making calculation on them is often waste of
resources.
12. Topic based method
M.R.Khadivi - Trending topics in Social Networks ComInSyS 12
o Topic detection methods
• BNgrams
▪ df-idft scores
• Topic Clustering
▪ Hierarchical clustering
▪ Apriori algorithm
▪ Gaussian mixture models
• Topic Ranking
▪ Maximum n-gram
▪ Weighted topic-length
13. Topic trending VS. Topic discovery
ComInSySM.R.Khadivi - Trending topics in Social Networks 13
14. Topic trending VS. Topic discovery
ComInSySM.R.Khadivi - Trending topics in Social Networks 14
15. Time series
ComInSySM.R.Khadivi - Trending topics in Social Networks 15
• is time-based
▪ if W consists of a sequence of fixed-length time units, where a variable
number of transactions may arrive within each time unit.
• is count-based
▪ if W is composed of a sequence of batches, where each batch consists of an
equal number of transactions.
⁉ Note that a count-based window can also be captured by a time-based
window by assuming that a uniform number of transactions arrive within each
time unit.
16. Time series – Cont.
ComInSySM.R.Khadivi - Trending topics in Social Networks 16
• A transaction data stream is a sequence of incoming transactions and an
excerpt of the stream is called a window
• Landmark window
• Damped window
• Sliding window
• Adaptive Window
17. Time series - Landmark Window
ComInSySM.R.Khadivi - Trending topics in Social Networks 17
18. Time series - Damped Window
ComInSySM.R.Khadivi - Trending topics in Social Networks 18
19. Time series - Fixed Sliding Window
ComInSySM.R.Khadivi - Trending topics in Social Networks 19
20. Time series - Adaptive Window
ComInSySM.R.Khadivi - Trending topics in Social Networks 20
21. Topical time series model
ComInSySM.R.Khadivi - Trending topics in Social Networks 21
• Occurrence Sequence
〈t,v,l〉 is an occurrence of l
For a topic l0,
𝑡𝑠𝑙0
= 𝑡0, 𝑣0, 𝑙0 , 𝑡1, 𝑣1, 𝑙0 , … (𝑡0 < 𝑡1 < ⋯)
22. Topical time series model
ComInSySM.R.Khadivi - Trending topics in Social Networks 22
• Popularity
A topic l's popularity: 𝜌𝑙: 𝑡 → 𝑅+
𝜌𝑙 𝑡 : the popularity value of l at time t
The density of occurrences of l at time t
Kernel density estimation
Window Size
occurrences
23. Topical time series model
ComInSySM.R.Khadivi - Trending topics in Social Networks 23
• The popularity sequence model
24. Social influence sequence
ComInSySM.R.Khadivi - Trending topics in Social Networks 24
• Topic influence
A topic l's influence: 𝜆𝑙: 𝑡 → 𝑅+
𝜆𝑙 𝑡 : the popularity value of l at time t
The density of occurrences of l at time t
Authority function
26. Influence with location factor
ComInSySM.R.Khadivi - Trending topics in Social Networks 26
• Location popularity
A topic l's influence: 𝜙𝑙: 𝑡 × 𝐹2
→ 𝑅+
𝜙𝑙 𝑡, 𝑝 : the popularity value of l at time t and location p
density of occurrences of l at time t and a specific location p.