RecSys
      • Sharing URL Link copied
      • /edit
      • View mode
        • Edit mode
        • View mode
        • Book mode
        • Slide mode
        Edit mode View mode Book mode Slide mode
      • Customize slides
      • Note Permission
      • Read
        • Owners
        • Signed-in users
        • Everyone
        Owners Signed-in users Everyone
      • Write
        • Owners
        • Signed-in users
        • Everyone
        Owners Signed-in users Everyone
      • Engagement control Commenting, Suggest edit, Emoji Reply
    • Invite by email
      Invitee

      This note has no invitees

    • Publish Note

      Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note No publishing access yet

      Your note will be visible on your profile and discoverable by anyone.
      Your note is now live.
      This note is visible on your profile and discoverable online.
      Everyone on the web can find and read all notes of this public team.

      Your account was recently created. Publishing will be available soon, allowing you to share notes on your public page and in search results.

      Your team account was recently created. Publishing will be available soon, allowing you to share notes on your public page and in search results.

      Explore these features while you wait
      Complete general settings
      Bookmark and like published notes
      Write a few more notes
      Complete general settings
      Write a few more notes
      See published notes
      Unpublish note
      Please check the box to agree to the Community Guidelines.
      View profile
    • Commenting
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
      • Everyone
    • Suggest edit
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
    • Emoji Reply
    • Enable
    • Versions and GitHub Sync
    • Note settings
    • Note Insights New
    • Engagement control
    • Make a copy
    • Transfer ownership
    • Delete this note
    • Insert from template
    • Import from
      • Dropbox
      • Google Drive
      • Gist
      • Clipboard
    • Export to
      • Dropbox
      • Google Drive
      • Gist
    • Download
      • Markdown
      • HTML
      • Raw HTML
Menu Note settings Note Insights Versions and GitHub Sync Sharing URL Help
Menu
Options
Engagement control Make a copy Transfer ownership Delete this note
Import from
Dropbox Google Drive Gist Clipboard
Export to
Dropbox Google Drive Gist
Download
Markdown HTML Raw HTML
Back
Sharing URL Link copied
/edit
View mode
  • Edit mode
  • View mode
  • Book mode
  • Slide mode
Edit mode View mode Book mode Slide mode
Customize slides
Note Permission
Read
Owners
  • Owners
  • Signed-in users
  • Everyone
Owners Signed-in users Everyone
Write
Owners
  • Owners
  • Signed-in users
  • Everyone
Owners Signed-in users Everyone
Engagement control Commenting, Suggest edit, Emoji Reply
  • Invite by email
    Invitee

    This note has no invitees

  • Publish Note

    Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note No publishing access yet

    Your note will be visible on your profile and discoverable by anyone.
    Your note is now live.
    This note is visible on your profile and discoverable online.
    Everyone on the web can find and read all notes of this public team.

    Your account was recently created. Publishing will be available soon, allowing you to share notes on your public page and in search results.

    Your team account was recently created. Publishing will be available soon, allowing you to share notes on your public page and in search results.

    Explore these features while you wait
    Complete general settings
    Bookmark and like published notes
    Write a few more notes
    Complete general settings
    Write a few more notes
    See published notes
    Unpublish note
    Please check the box to agree to the Community Guidelines.
    View profile
    Engagement control
    Commenting
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    • Everyone
    Suggest edit
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    Emoji Reply
    Enable
    Import from Dropbox Google Drive Gist Clipboard
       Owned this note    Owned this note      
    Published Linked with GitHub
    • Any changes
      Be notified of any changes
    • Mention me
      Be notified of mention me
    • Unsubscribe
    # Cap. 5: Content Based Recommender Systems ### What distinguishes content-based recommender systems from collaborative filtering methods in terms of data usage and recommendations? I RS Content-based, a differenza dei Collaborative Filtering, non tengono conto dei rating forniti da altri utenti, ma solo dallo storico dell'utente (per il quale si sta facendo la raccomandazione) e dalle caratteristiche degli item, per cui non è necessaria la matrice dei rating (così come vista per CFRS). ### In content-based recommender systems, what are the two main sources of data used for making recommendations? How do they help address certain challenges? > Alternatives: > - Explain how content-based recommender systems leverage item attributes to make recommendations and why they are particularly useful for scenarios with descriptive attributes. > - How do content-based systems address the "cold-start problem" for new users? > - Give examples of scenarios where content-based recommendations are effective, such as movie recommendations, web page recommendations, and music recommendations. Le fonti di dati principali usate nei CBRS sono: - le descrizioni degli item (che quanto più ricche di feature sono, meglio è) - un profilo dei gusti dell'utente (che può essere uno storico, dei rating, un tema verso il quale ha dimostrato interesse) I CBRS possono essere particolarmente utili: - per risolvere il problema del cold start (in quanto richiedono meno dati dal punto di vista del profilo utente); - in tutti quei contesti dove è possibile estrarre tanto contenuto dagli item (testi, musica, ecc.) o dove sono presenti molti dati non strutturati; - Hanno un forte collegamento con i *Knowledge-based Rec Sys* ### Describe the components of a content-based recommender system, including preprocessing, feature extraction, learning user profiles, and filtering. > Alternative: > - What is feature extraction, and how does it play a role in content-based systems? Provide examples of how features are extracted from different domains. > - What type of data is used to build user profiles in content-based systems, and how is this data collected? > - In content-based systems, why is feature representation and cleaning an important step? What preprocessing steps are commonly applied in this phase? I componenti di un CBRS sono quelli che ci si può aspettare di trovare in un task di data science. 1) Innanzitutto serve un dataset con quanti più dati (strutturati e non) sugli item. 2) Si svolge una fase di pre-processing finalizzata ad estrarre **feature discriminanti** che permettano di distinguere gli item tra di loro, che verranno pesate più o meno nel modello. Per ciascuna feature si dovrà trovare una rappresentazione adeguata (per alcune le si terrà "native", per quelle testuali tornano utili rappresentazioni come BOW, ecc.). 3) Il pre-processing include, specie per i dati non strutturati, anche operazioni di pulizia (ad es. per il testo, stemming, stopwords removal, phrase extraction, ecc.) 4) L'obiettivo finale è arrivare ad un vettore di feature. 5) Dal punto del profilo utente invece è importante catturare le sue iterazioni con il sistema, quindi: - ratings - like e dislike - feedback impliciti - ecc. 6) A volte le feature che si ricavano sono anche fin troppe, quindi può essere utile fare una *selezione* e una *pesatura*. Per fare ciò tornano utili tecniche come: - l'indice di Gini - l'entropia e l'information Gain Tutte queste tecniche sono utili a capire "quanto una feature è discriminante" per i rating dell'utente (voglio selezionare feature "predittive" del voto) 7) Dati gli item rappresentati come vettori di feature e un set di item valutati, tramite un algoritmo di machine learning (anche molto naive come K-Nearest Neighbors) posso prevedere i rating degli item mancanti. ### Explain the process of collecting user likes and dislikes and how they are converted into ratings in content-based systems. Un paio di esempi: - raccogliere feedback espliciti come like, dislike, rating, ecc. sottoforma di dato binario, discreto, etc. - raccogliere feedback impliciti come click, buy, azioni varie, ecc. - raccogliere e analizzare feedback testuali (ad es., fare sentimental analysis sui commenti, ecc.) - raccogliere esempi "positivi e negativi" di ciò che l'utente vuole ### What is the purpose of supervised feature selection and weighting in content-based systems? How does it impact the vector-space representation of keywords? > Alternative: > - Describe the Gini Index and its application in feature selection. How does it measure feature discriminative power? > - Explain the concept of entropy and how it is used for feature selection in content-based systems. How does it differ from the Gini Index? Specie quando si analizzano dati non strutturati le feature possono essere tante; è necessario quindi fare una selezione di quelle più utili a discriminare gli item (dal punto di vista dell'utente). 1) Prendo tutti gli item valutati dall'utente (con le relative valutazioni) 2) Prendi le singole feature e ne calcolo il contenuto informativo (ad esempio con l'indice di Gini e/o con l'Information Gain) rispetto alle valutazioni degli item in cui la feature è presente. $$Gini(w)=1-\sum_{i=0}^t(p_i(w))^2$$ In pratica, dati gli item aventi una feature, considero la distribuzione di probabilità $p_1(w),..., p_t(w)$ dei rating assegnati dall'utente (dove $1,...,t$ sono tutti i possibili valori) ai documenti aventi una certa feature. Sono quindi interessato a selezionare feature per le quali questa distribuzione è disomogenea (le quali hanno alto potere predittivo rispetto al rating che l'utente darà). > NOTA: $p_i(w) = P(i | w)$, quindi la frazione di item valutati $i$ tra quelli contenenti la feature $w$. Si noti che ogni utente quindi avrà feature personalizzate in base a cosa ritiene polarizzante (in alternativa all'approccio classico di data science dove si premiano feature utili a distinguere gli oggetti tra di loro). Da evidenze sperimentali emerge che di solito va bene tenere tra le 50 e le 300 feature. Vengono poi considerati solo gli item che contengono le feature scelte. ### How do content-based systems learn user profiles, and how is this process related to classification and regression modeling? > Alternative: > - Describe the Nearest Neighbor Classification technique in content-based systems. How does it predict ratings for items in the test set? > - Explain the concept of user-specific models in content-based systems and how they are constructed. L'apprendimento di un profilo utente è un problema di machine learning (regressione/classificazione), che può essere risolto anche con un algoritmo naive come **Nearest Neighbor Classification**: - c'è un set di training (item per i quali l'utente ha dato un rating) e uno di test per i quali devo stimare i rating ($D_L, D_U$) - data una coppia di documenti (vettori $X,Y$), si può definire una misura di similarità (e.g., $Cosine(X,Y) = \frac{X^TY}{\sqrt{||X||_2} \sqrt{||Y||_2}}$) - per ogni documento del test set, prendo i k documenti più vicini nel training e ci faccio la media dei rating per fare previsione (la media può essere pesata con la similarità) - la complessità risultante è $O(|D_L| * |D_U|)$ - da un certo punto di vista "si sta facendo clustering" ### What is the advantage of content-based recommendations in terms of providing explanations for the recommendations made? How do content-based recommendations benefit from the interpretability of predictions? I sistemi content-based sono capaci di fornire spiegazioni in quanto: - le raccomandazioni avvengo su base di feature reali selezionate - "ti suggerisco X perché contiene A,B e C" - tali feature sono indipendenti dagli altri utenti (e quindi ci sono meno problemi di privacy) - quando genero le raccomandazioni so anche quali documenti sono stati considerati per stimare il rating - "ti suggerisco X perché assomiglia ad Y e Z" ### Compare the advantages and disadvantages of content-based recommender systems in contrast to collaborative filtering methods. > Alternative: > - How do content-based recommender systems perform in terms of novelty and serendipity compared to collaborative filtering methods? How do content-based methods address the diversity and novelty of recommended items compared to collaborative filtering? > - What is the "cold start" problem in the context of content-based recommender systems? I vantaggi sono: - l'explainability, - facile da implementare anche con componenti terzi - l'indipendenza dalla popolarità (e quindi la capacità di consigliare nuovi item non ancora valutati dagli altri utenti) - richiede meno dati rispetto ai modelli CF I svantaggi invece sono: - limitata esplorazione (problemi di novelty e serendipity) - anche questi modelli soffrono il problema del cold start ### How can content-based models be used for collaborative filtering? Provide examples of how the data can be mapped to support collaborative filtering. > Alternative: > - How does the mapping of user and item interactions in a content-based model approximate item-based neighborhood models for collaborative filtering? > - Can content-based models be used in combination with off-the-shelf content-based algorithms to support collaborative filtering? If yes, how? Un modello CB può essere adattato per fare anche CF aggiungendo agli item delle feature contenenti i rating di ciascun utente. A questo punto l'algoritmo di k-nearest neighborhood diventa lo stesso.

    Import from clipboard

    Paste your markdown or webpage here...

    Advanced permission required

    Your current role can only read. Ask the system administrator to acquire write and comment permission.

    This team is disabled

    Sorry, this team is disabled. You can't edit this note.

    This note is locked

    Sorry, only owner can edit this note.

    Reach the limit

    Sorry, you've reached the max length this note can be.
    Please reduce the content or divide it to more notes, thank you!

    Import from Gist

    Import from Snippet

    or

    Export to Snippet

    Are you sure?

    Do you really want to delete this note?
    All users will lose their connection.

    Create a note from template

    Create a note from template

    Oops...
    This template has been removed or transferred.
    Upgrade
    All
    • All
    • Team
    No template.

    Create a template

    Upgrade

    Delete template

    Do you really want to delete this template?
    Turn this template into a regular note and keep its content, versions, and comments.

    This page need refresh

    You have an incompatible client version.
    Refresh to update.
    New version available!
    See releases notes here
    Refresh to enjoy new features.
    Your user state has changed.
    Refresh to load new user state.

    Sign in

    Forgot password
    or
    Sign in via Facebook Sign in via X(Twitter) Sign in via GitHub Sign in via Dropbox Sign in with Wallet
    Wallet ( )
    Connect another wallet

    New to HackMD? Sign up

    By signing in, you agree to our terms of service.

    Help

    • English
    • 中文
    • Français
    • Deutsch
    • 日本語
    • Español
    • Català
    • Ελληνικά
    • Português
    • italiano
    • Türkçe
    • Русский
    • Nederlands
    • hrvatski jezik
    • język polski
    • Українська
    • हिन्दी
    • svenska
    • Esperanto
    • dansk

    Documents

    Help & Tutorial

    How to use Book mode

    Slide Example

    API Docs

    Edit in VSCode

    Install browser extension

    Contacts

    Feedback

    Discord

    Send us email

    Resources

    Releases

    Pricing

    Blog

    Policy

    Terms

    Privacy

    Cheatsheet

    Syntax Example Reference
    # Header Header 基本排版
    - Unordered List
    • Unordered List
    1. Ordered List
    1. Ordered List
    - [ ] Todo List
    • Todo List
    > Blockquote
    Blockquote
    **Bold font** Bold font
    *Italics font* Italics font
    ~~Strikethrough~~ Strikethrough
    19^th^ 19th
    H~2~O H2O
    ++Inserted text++ Inserted text
    ==Marked text== Marked text
    [link text](https:// "title") Link
    ![image alt](https:// "title") Image
    `Code` Code 在筆記中貼入程式碼
    ```javascript
    var i = 0;
    ```
    var i = 0;
    :smile: :smile: Emoji list
    {%youtube youtube_id %} Externals
    $L^aT_eX$ LaTeX
    :::info
    This is a alert area.
    :::

    This is a alert area.

    Versions and GitHub Sync
    Get Full History Access

    • Edit version name
    • Delete

    revision author avatar     named on  

    More Less

    Note content is identical to the latest version.
    Compare
      Choose a version
      No search result
      Version not found
    Sign in to link this note to GitHub
    Learn more
    This note is not linked with GitHub
     

    Feedback

    Submission failed, please try again

    Thanks for your support.

    On a scale of 0-10, how likely is it that you would recommend HackMD to your friends, family or business associates?

    Please give us some advice and help us improve HackMD.

     

    Thanks for your feedback

    Remove version name

    Do you want to remove this version name and description?

    Transfer ownership

    Transfer to
      Warning: is a public team. If you transfer note to this team, everyone on the web can find and read this note.

        Link with GitHub

        Please authorize HackMD on GitHub
        • Please sign in to GitHub and install the HackMD app on your GitHub repo.
        • HackMD links with GitHub through a GitHub App. You can choose which repo to install our App.
        Learn more  Sign in to GitHub

        Push the note to GitHub Push to GitHub Pull a file from GitHub

          Authorize again
         

        Choose which file to push to

        Select repo
        Refresh Authorize more repos
        Select branch
        Select file
        Select branch
        Choose version(s) to push
        • Save a new version and push
        • Choose from existing versions
        Include title and tags
        Available push count

        Pull from GitHub

         
        File from GitHub
        File from HackMD

        GitHub Link Settings

        File linked

        Linked by
        File path
        Last synced branch
        Available push count

        Danger Zone

        Unlink
        You will no longer receive notification when GitHub file changes after unlink.

        Syncing

        Push failed

        Push successfully