Introduction To Recommender Systems an der Universität Bern | Karteikarten & Zusammenfassungen

# Lernmaterialien für Introduction To Recommender Systems an der Universität Bern

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Introduction To Recommender Systems Kurs an der Universität Bern zu.

TESTE DEIN WISSEN

Lösung anzeigen
TESTE DEIN WISSEN

• Set of users can be small
• Good for finding subsitutes
• Good for finding the right product
• Good explainability

• Unlikely to find surprising connections
• Harder to find complements (rather than substitutes)
Lösung ausblenden
TESTE DEIN WISSEN

What is the difference between predictions and recommendations?

Lösung anzeigen
TESTE DEIN WISSEN

In the context of Recommender Systems, a prediction represents the anticipated opinion of a certain user for a given item. The predicted rating should have the same numerical scale as the real ratings.

A recommendation, on the other hand, is a list (subset) of all the items. The items in the list are chosen in a way that the user will probably like them the most of all the items in the system.

Lösung ausblenden
TESTE DEIN WISSEN

What are some methods to reduce the sparsity of the user-item matrix in CBF?

Lösung anzeigen
TESTE DEIN WISSEN
• Default Voting: insert default voting for unrated items, which corresponds to neutral or somewhat negative preference.
• Preprocessing using Averages: for each user, assign his or her mean rating score to the unobserved items.
• Filterbots: use (intelligent) automated rating agents to rate new items.
• Dimensionality Reduction: use dimensionality reduction techniques like SVD. This may also solve the synonymy problem since the latent representation of two items with different names might be similar.
Lösung ausblenden
TESTE DEIN WISSEN

What are two different ways of forming a neighborhood of users in CBF?

Lösung anzeigen
TESTE DEIN WISSEN

We can either just pick the useres that have the highest similarity with the active user or we can incrementally add a new user to the neighborhood, calculate the new centroid of this neighborhood and proceed by selecting the next user that is closest to this centroid.

Lösung ausblenden
TESTE DEIN WISSEN

Name two different Top-N Recommendation strategies in CBF.

Lösung anzeigen
TESTE DEIN WISSEN
• Recommend the N most frequent items that have been rated in the neighborhood of the active user but not by the active user itself.
• Make use of association rules and choose the ones with the highest confidence.
Lösung ausblenden
TESTE DEIN WISSEN

What are two assumptions made in CBF methods?

Lösung anzeigen
TESTE DEIN WISSEN
• User tastes are stable
• For a given domain (e.g. movies): if two users agree on one type of item, thy also agree on other types of items
Lösung ausblenden
TESTE DEIN WISSEN

How many neighbors k should we consider in user-user CBF?

Lösung anzeigen
TESTE DEIN WISSEN

The lecture slides are somewhat ambivalent regarding this question. In general, seems to be often somewhere in between 25 and 100 in many domains, but it might be better to start with a value between 20 or 30 and 50. For movie ratings, Herlocker et al. found k = 20 to be a good value.

Lösung ausblenden
TESTE DEIN WISSEN

Could we also make use of users that rate items very differently from the active user for predicting new items?

Lösung anzeigen
TESTE DEIN WISSEN

Yes. For example, if a user always rates things very low that the active user rates very high, we can infer that their tastes are somewhat orthogonal. We could use the bad ratings of this user to predict items that the active user is going to like.

Lösung ausblenden
TESTE DEIN WISSEN

What are some similaritiy metrics that can be used in user-user CBF?

Lösung anzeigen
TESTE DEIN WISSEN
• Pearson Correlation
• Constrained Pearson Correlation
• Spearman Rank Correlation
• Cosine Similarity
Lösung ausblenden
TESTE DEIN WISSEN

What is the difference between Pearson Correlation and Constrained Pearson Correlation?

Lösung anzeigen
TESTE DEIN WISSEN

In the Constrained Pearson Correlation, we use a neutral value instead of the mean rating of each user. This neutral value should denote neither agreement nor disagreement.

Lösung ausblenden
TESTE DEIN WISSEN

What are the advantages of item-item CBF over user-user CBF?

Lösung anzeigen
TESTE DEIN WISSEN
• Item-item similarity is fairly stable.
• We can precompute the item-item similarity (since it is stable)
• Faster recommendations for systems with a lot of users.

Lösung ausblenden
TESTE DEIN WISSEN

How does TF-IDF apply to CBF?

Lösung anzeigen
TESTE DEIN WISSEN

The TF-IDF concept can be used to create a profile of a document/object. A movie could be described as a weighted vector of its tags. These TF-IDF profiles can be combined with ratings to create user profiles, and then matched against future documents.

Lösung ausblenden
• 51525 Karteikarten
• 810 Studierende
• 22 Lernmaterialien

## Beispielhafte Karteikarten für deinen Introduction To Recommender Systems Kurs an der Universität Bern - von Kommilitonen auf StudySmarter erstellt!

Q:

A:

• Set of users can be small
• Good for finding subsitutes
• Good for finding the right product
• Good explainability

• Unlikely to find surprising connections
• Harder to find complements (rather than substitutes)
Q:

What is the difference between predictions and recommendations?

A:

In the context of Recommender Systems, a prediction represents the anticipated opinion of a certain user for a given item. The predicted rating should have the same numerical scale as the real ratings.

A recommendation, on the other hand, is a list (subset) of all the items. The items in the list are chosen in a way that the user will probably like them the most of all the items in the system.

Q:

What are some methods to reduce the sparsity of the user-item matrix in CBF?

A:
• Default Voting: insert default voting for unrated items, which corresponds to neutral or somewhat negative preference.
• Preprocessing using Averages: for each user, assign his or her mean rating score to the unobserved items.
• Filterbots: use (intelligent) automated rating agents to rate new items.
• Dimensionality Reduction: use dimensionality reduction techniques like SVD. This may also solve the synonymy problem since the latent representation of two items with different names might be similar.
Q:

What are two different ways of forming a neighborhood of users in CBF?

A:

We can either just pick the useres that have the highest similarity with the active user or we can incrementally add a new user to the neighborhood, calculate the new centroid of this neighborhood and proceed by selecting the next user that is closest to this centroid.

Q:

Name two different Top-N Recommendation strategies in CBF.

A:
• Recommend the N most frequent items that have been rated in the neighborhood of the active user but not by the active user itself.
• Make use of association rules and choose the ones with the highest confidence.
Q:

What are two assumptions made in CBF methods?

A:
• User tastes are stable
• For a given domain (e.g. movies): if two users agree on one type of item, thy also agree on other types of items
Q:

How many neighbors k should we consider in user-user CBF?

A:

The lecture slides are somewhat ambivalent regarding this question. In general, seems to be often somewhere in between 25 and 100 in many domains, but it might be better to start with a value between 20 or 30 and 50. For movie ratings, Herlocker et al. found k = 20 to be a good value.

Q:

Could we also make use of users that rate items very differently from the active user for predicting new items?

A:

Yes. For example, if a user always rates things very low that the active user rates very high, we can infer that their tastes are somewhat orthogonal. We could use the bad ratings of this user to predict items that the active user is going to like.

Q:

What are some similaritiy metrics that can be used in user-user CBF?

A:
• Pearson Correlation
• Constrained Pearson Correlation
• Spearman Rank Correlation
• Cosine Similarity
Q:

What is the difference between Pearson Correlation and Constrained Pearson Correlation?

A:

In the Constrained Pearson Correlation, we use a neutral value instead of the mean rating of each user. This neutral value should denote neither agreement nor disagreement.

Q:

What are the advantages of item-item CBF over user-user CBF?

A:
• Item-item similarity is fairly stable.
• We can precompute the item-item similarity (since it is stable)
• Faster recommendations for systems with a lot of users.

Q:

How does TF-IDF apply to CBF?

A:

The TF-IDF concept can be used to create a profile of a document/object. A movie could be described as a weighted vector of its tags. These TF-IDF profiles can be combined with ratings to create user profiles, and then matched against future documents.

### Erstelle und finde Lernmaterialien auf StudySmarter.

Greife kostenlos auf tausende geteilte Karteikarten, Zusammenfassungen, Altklausuren und mehr zu.

## Das sind die beliebtesten Introduction To Recommender Systems Kurse im gesamten StudySmarter Universum

##### Introduction to computers

Nigerian Turkish Nile University

##### Introduction to communication

Nigerian Turkish Nile University

##### Introduction to Education

Baptist Bible College

##### B3 Integrated Production Systems

Technische Hochschule Nürnberg