Collab at TU Kaiserslautern | Flashcards & Summaries

Select your language

Suggested languages for you:
Log In Start studying!

Lernmaterialien für Collab an der TU Kaiserslautern

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Collab Kurs an der TU Kaiserslautern zu.

TESTE DEIN WISSEN

blindness while moving the eye

Lösung anzeigen
TESTE DEIN WISSEN

saccadic suppression

Lösung ausblenden
TESTE DEIN WISSEN

Grundlagen von Recommender Systemen 

Users

Lösung anzeigen
TESTE DEIN WISSEN

Users - Features 

  • Aufbau: Name des Features + Ausprägung 
  • Klassische Features: 
    •  Demografisch: Alter, Geschlecht, …  
    • Ort: Geburtsort, aktuelle Adresse, … 
    • Psychologische Features: Interessen, Hobbies, Musikgeschmack …
Lösung ausblenden
TESTE DEIN WISSEN

semantic or contextual closeness / contextual search

Lösung anzeigen
TESTE DEIN WISSEN

● This idea is based on the assumption that two terms, which frequently occur in the same context, are associatively related to each other 

● As a consequence we statistically measure the co-occurrences of two words, which is called contextual search

Lösung ausblenden
TESTE DEIN WISSEN

Eye tracking technical overview

Lösung anzeigen
TESTE DEIN WISSEN

● Gaze data is collected using either a remote or head-mounted eye-tracker 

● A (infrared) light source is directed toward the eye 

● Camera tracks the reflection of the light source along with visible ocular features 

● Data is used to extrapolate eye rotation as well as the direction of gaze 

● Visual path is analyzed across an interface and transformed into a set of pixel coordinates, i.e. 

○ which features are seen, when a particular feature captures attention, how quick is the eye moving, what content is overlooked, etc.

Lösung ausblenden
TESTE DEIN WISSEN

Recommender Systems 

Why do we need filtering?

Lösung anzeigen
TESTE DEIN WISSEN

● Catalogue size (e.g. iTunes: 6 Million songs in 2009!) 

● Several million new songs each day 

● Long Tail Effect: “Music Overload” generates demand for filtering, recommendation, personalization

Lösung ausblenden
TESTE DEIN WISSEN

Controlled vocabularies / thesaur

Lösung anzeigen
TESTE DEIN WISSEN

Möglichkeiten das Ergebnis der Suche zu verbessern: 

→ Metadaten: Entweder manuell oder automatisiert zu Dokumenten hinzugefügt 

● using it to “expand the semantic nature of a query” as “data about data” 

● using it to annotate the documents in a repository with useful information about e.g. their content, purpose and origin 

● improving classification, ranking and relevance by exploiting the content of these “tags” and use the information stored in them

Lösung ausblenden
TESTE DEIN WISSEN

How can we evaluate a search (information retrieval) system

Lösung anzeigen
TESTE DEIN WISSEN

Precision + Recall // F-Measure

Lösung ausblenden
TESTE DEIN WISSEN

Inverted index with term weights

Lösung anzeigen
TESTE DEIN WISSEN

A term vector may contain just the value “true” as an indicator that a word is part of a document or may show a weight, e.g. according to the frequency of their occurrence in a document (Assign weights instead of boolean values) 

→ The importance of an index term can be expressed by its weight x 

→ Easy processing of a query, e.g. by summarizing the weights of the terms of a conjunction

Lösung ausblenden
TESTE DEIN WISSEN

Relevance Feedback (or Pseudo Relevance Feedback)

Lösung anzeigen
TESTE DEIN WISSEN

The idea of the approach is to involve the user‘s preference in the retrieval process in order to improve the ranking of the results → i.e. the user gives feedback on the relevance of documents in an initial set of results.

Lösung ausblenden
TESTE DEIN WISSEN

5 Steps of Relevance Feedback

Lösung anzeigen
TESTE DEIN WISSEN

1. The user issues a (short, simple) query 

2. The system returns an first result list retrieved from a given corpus. 

3. The user labels some returned documents as relevant or non relevant. 

4. The system computes a better representation of the information need based on the user feedback. 

5. The system displays a revised set of retrieval results.

Lösung ausblenden
TESTE DEIN WISSEN

Gaze-Based Self-Condence Estimation Research Hypotheses: 

● RH1: Questions answered correctly without confidence tend to be forgotten compared to knowledge with confidence

● RH2: Questions answered incorrectly with confidence tend to be mistaken again compared to wrong knowledge without confidence 

● RH3: Estimating self-confidence from learning behaviors and giving feedback (e.g., adding questions to a review list, highlighting them while reviewing) avoids such scenarios.

Lösung anzeigen
TESTE DEIN WISSEN

● RH1: Questions answered correctly without confidence tend to be forgotten compared to knowledge with confidence. → True 

● RH2: Questions answered incorrectly with confidence tend to be mistaken again compared to wrong knowledge without confidence. → Not always true 

● RH3: Estimating self-confidence from learning behaviors and giving feedback (e.g., adding questions to a review list, highlighting them while reviewing) avoids such scenarios. → True

Lösung ausblenden
TESTE DEIN WISSEN

Recommender systems 

Definition

Lösung anzeigen
TESTE DEIN WISSEN

A recommender system R is a program which recommends to user Ui a set of items Ij according to his preferences.

Lösung ausblenden
  • 45410 Karteikarten
  • 1065 Studierende
  • 78 Lernmaterialien

Beispielhafte Karteikarten für deinen Collab Kurs an der TU Kaiserslautern - von Kommilitonen auf StudySmarter erstellt!

Q:

blindness while moving the eye

A:

saccadic suppression

Q:

Grundlagen von Recommender Systemen 

Users

A:

Users - Features 

  • Aufbau: Name des Features + Ausprägung 
  • Klassische Features: 
    •  Demografisch: Alter, Geschlecht, …  
    • Ort: Geburtsort, aktuelle Adresse, … 
    • Psychologische Features: Interessen, Hobbies, Musikgeschmack …
Q:

semantic or contextual closeness / contextual search

A:

● This idea is based on the assumption that two terms, which frequently occur in the same context, are associatively related to each other 

● As a consequence we statistically measure the co-occurrences of two words, which is called contextual search

Q:

Eye tracking technical overview

A:

● Gaze data is collected using either a remote or head-mounted eye-tracker 

● A (infrared) light source is directed toward the eye 

● Camera tracks the reflection of the light source along with visible ocular features 

● Data is used to extrapolate eye rotation as well as the direction of gaze 

● Visual path is analyzed across an interface and transformed into a set of pixel coordinates, i.e. 

○ which features are seen, when a particular feature captures attention, how quick is the eye moving, what content is overlooked, etc.

Q:

Recommender Systems 

Why do we need filtering?

A:

● Catalogue size (e.g. iTunes: 6 Million songs in 2009!) 

● Several million new songs each day 

● Long Tail Effect: “Music Overload” generates demand for filtering, recommendation, personalization

Mehr Karteikarten anzeigen
Q:

Controlled vocabularies / thesaur

A:

Möglichkeiten das Ergebnis der Suche zu verbessern: 

→ Metadaten: Entweder manuell oder automatisiert zu Dokumenten hinzugefügt 

● using it to “expand the semantic nature of a query” as “data about data” 

● using it to annotate the documents in a repository with useful information about e.g. their content, purpose and origin 

● improving classification, ranking and relevance by exploiting the content of these “tags” and use the information stored in them

Q:

How can we evaluate a search (information retrieval) system

A:

Precision + Recall // F-Measure

Q:

Inverted index with term weights

A:

A term vector may contain just the value “true” as an indicator that a word is part of a document or may show a weight, e.g. according to the frequency of their occurrence in a document (Assign weights instead of boolean values) 

→ The importance of an index term can be expressed by its weight x 

→ Easy processing of a query, e.g. by summarizing the weights of the terms of a conjunction

Q:

Relevance Feedback (or Pseudo Relevance Feedback)

A:

The idea of the approach is to involve the user‘s preference in the retrieval process in order to improve the ranking of the results → i.e. the user gives feedback on the relevance of documents in an initial set of results.

Q:

5 Steps of Relevance Feedback

A:

1. The user issues a (short, simple) query 

2. The system returns an first result list retrieved from a given corpus. 

3. The user labels some returned documents as relevant or non relevant. 

4. The system computes a better representation of the information need based on the user feedback. 

5. The system displays a revised set of retrieval results.

Q:

Gaze-Based Self-Condence Estimation Research Hypotheses: 

● RH1: Questions answered correctly without confidence tend to be forgotten compared to knowledge with confidence

● RH2: Questions answered incorrectly with confidence tend to be mistaken again compared to wrong knowledge without confidence 

● RH3: Estimating self-confidence from learning behaviors and giving feedback (e.g., adding questions to a review list, highlighting them while reviewing) avoids such scenarios.

A:

● RH1: Questions answered correctly without confidence tend to be forgotten compared to knowledge with confidence. → True 

● RH2: Questions answered incorrectly with confidence tend to be mistaken again compared to wrong knowledge without confidence. → Not always true 

● RH3: Estimating self-confidence from learning behaviors and giving feedback (e.g., adding questions to a review list, highlighting them while reviewing) avoids such scenarios. → True

Q:

Recommender systems 

Definition

A:

A recommender system R is a program which recommends to user Ui a set of items Ij according to his preferences.

Collab

Erstelle und finde Lernmaterialien auf StudySmarter.

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

Jetzt loslegen

Das sind die beliebtesten Collab Kurse im gesamten StudySmarter Universum

colloq 1

Medical University Varna

Zum Kurs
Colloquium 1

Medical University Plovdiv

Zum Kurs
Columna

Universidad de Santiago de Chile

Zum Kurs
vene collo

University of Bari

Zum Kurs
CCLAB

Saint Louis College

Zum Kurs

Die all-in-one Lernapp für Studierende

Greife auf Millionen geteilter Lernmaterialien der StudySmarter Community zu
Kostenlos anmelden Collab
Erstelle Karteikarten und Zusammenfassungen mit den StudySmarter Tools
Kostenlos loslegen Collab