Breve introduzione ai Recommender Systems e al problema della diversità dei suggerimenti, con utili riferimenti bibliografici.
Slide usate durante il talk http://www.eventbrite.it/e/biglietti-club-degli-sviluppatori-puglia-recommender-systems-9384077027?aff=eorg
The concept of combined geothermal energy and Thermal Active Building Systems
(TABS), known as GEOTABS, has been developed the last years and has known great
success. The energy saving potential is substantial, starting from 20% and going up to
70%.For a system like this though, to reach its maximum potential of energy savings,
professional design, control and installation, combined with a product of highest quality,
must be combined from feasibility stage to final building-in-operation stage. The latter
means that experienced and professional partners are considered crucial for system
optimization
Lifting the lid on Business Rules - Robin Grace IndigoCube
Understanding Business Rules is key to understanding business analysis. Business Analysts must have an in-depth understanding of what business rules are and how to identify them accurately and completely when performing their business analysis tasks. Failing to accurately identify a business rule will result in a software defect that may be very costly to fix.
In this seminar, Robin Grace will explore the relationship between business rules and requirements, outlining a systematic approach to be used for the identification of business rules.
The concept of combined geothermal energy and Thermal Active Building Systems
(TABS), known as GEOTABS, has been developed the last years and has known great
success. The energy saving potential is substantial, starting from 20% and going up to
70%.For a system like this though, to reach its maximum potential of energy savings,
professional design, control and installation, combined with a product of highest quality,
must be combined from feasibility stage to final building-in-operation stage. The latter
means that experienced and professional partners are considered crucial for system
optimization
Lifting the lid on Business Rules - Robin Grace IndigoCube
Understanding Business Rules is key to understanding business analysis. Business Analysts must have an in-depth understanding of what business rules are and how to identify them accurately and completely when performing their business analysis tasks. Failing to accurately identify a business rule will result in a software defect that may be very costly to fix.
In this seminar, Robin Grace will explore the relationship between business rules and requirements, outlining a systematic approach to be used for the identification of business rules.
IndigoCube the agile enterprise: moving beyond scrum by JacoViljoenIndigoCube
To stay relevant in a world of accelerating change, business executives are increasingly striving for greater business agility.
To achieve this, the modern enterprise faces challenges such as:
• Increased responsiveness to market demands,
• Managing business agility at the portfolio and program level,
• Aligning business and IT agility,
• Extending software development agility to the greater application life cycle,
• Scaling agile practices so that it perpetuates throughout the organisation,
• Enabling agility using DevOps toolsets that significantly enhance productivity and speeds up delivery.
Join Jaco Viljoen, Principal consultant for Agile Software Development at IndigoCube and hear about the latest thinking in scaling agile to the enterprise and learn how to address these problems. Furthermore, Viljoen will discuss the state of agile today, agile frameworks for the agile enterprise, enabling DevOps toolsets, and how it all comes together to facilitate business agility.
With many organisations re-thinking the execution of their innovation lifecycles in an attempt to gain better productivity, some of the key questions that keep recurring are:
• When does a BA get allocated to a new business initiative?
• When does the business initiative become a project and require some form of project management?
• How does enterprise analysis fit into the systems development lifecycle?
• Who creates a business case?
• Who is assigned first: PM or BA?
Robin Grace, a business analysis principal consultants at IndigoCube, contributor to an IIBA white paper, CBAP, and author of Aligning Business Analysis, Assessing business analysis from a results focus, tells all.
"Challenges Faced by Testers Working on Agile Teams" by Aldo RallIndigoCube
"Challenges Faced by Testers Working on Agile Teams" by Aldo Rall
As a tester, moving into an Agile team can be frustrating and difficult. Often times leaving testers disillusioned and projects suffering due to a lack of quality.
In this talk, Aldo Rall will be looking at the typical challenges that testers face when moving into the Agile world, and touch on some key points that needs consideration for testers to successfully adapt in this new and often strange world called Agile.
Presentazione delle diverse tipologia di revisione: tradizionale, revisione sistematica e meta-analisi, revisione sistematiche qualitative e metasintesi, revisioni metodi misti
Corso di Laurea in Logopedia
Corso di Laurea in Infermieristica
AA 2019/2020
Machine Learning: strategie di collaborative filtering nelle piattaforme onli...Federico Panini
Machine Learning: strategie di collaborative filtering nelle piattaforme online. Un caso concreto di applicazione nel mondo SaaS e E-commerce, il caso Fazland.com.
IndigoCube the agile enterprise: moving beyond scrum by JacoViljoenIndigoCube
To stay relevant in a world of accelerating change, business executives are increasingly striving for greater business agility.
To achieve this, the modern enterprise faces challenges such as:
• Increased responsiveness to market demands,
• Managing business agility at the portfolio and program level,
• Aligning business and IT agility,
• Extending software development agility to the greater application life cycle,
• Scaling agile practices so that it perpetuates throughout the organisation,
• Enabling agility using DevOps toolsets that significantly enhance productivity and speeds up delivery.
Join Jaco Viljoen, Principal consultant for Agile Software Development at IndigoCube and hear about the latest thinking in scaling agile to the enterprise and learn how to address these problems. Furthermore, Viljoen will discuss the state of agile today, agile frameworks for the agile enterprise, enabling DevOps toolsets, and how it all comes together to facilitate business agility.
With many organisations re-thinking the execution of their innovation lifecycles in an attempt to gain better productivity, some of the key questions that keep recurring are:
• When does a BA get allocated to a new business initiative?
• When does the business initiative become a project and require some form of project management?
• How does enterprise analysis fit into the systems development lifecycle?
• Who creates a business case?
• Who is assigned first: PM or BA?
Robin Grace, a business analysis principal consultants at IndigoCube, contributor to an IIBA white paper, CBAP, and author of Aligning Business Analysis, Assessing business analysis from a results focus, tells all.
"Challenges Faced by Testers Working on Agile Teams" by Aldo RallIndigoCube
"Challenges Faced by Testers Working on Agile Teams" by Aldo Rall
As a tester, moving into an Agile team can be frustrating and difficult. Often times leaving testers disillusioned and projects suffering due to a lack of quality.
In this talk, Aldo Rall will be looking at the typical challenges that testers face when moving into the Agile world, and touch on some key points that needs consideration for testers to successfully adapt in this new and often strange world called Agile.
Presentazione delle diverse tipologia di revisione: tradizionale, revisione sistematica e meta-analisi, revisione sistematiche qualitative e metasintesi, revisioni metodi misti
Corso di Laurea in Logopedia
Corso di Laurea in Infermieristica
AA 2019/2020
Machine Learning: strategie di collaborative filtering nelle piattaforme onli...Federico Panini
Machine Learning: strategie di collaborative filtering nelle piattaforme online. Un caso concreto di applicazione nel mondo SaaS e E-commerce, il caso Fazland.com.
Adaptive choice based conjoint analysis vs full profile conjoint analysisTarget Research
Per distinguersi nel contesto competitivo è cruciale per un’azienda conoscere, sapersi adeguare e allo stesso tempo soddisfare le esigenze dei propri consumatori attuali e potenziali. La comprensione del comportamento del consumatore diventa così un presupposto necessario per creare un prodotto che rientri nelle preferenze dei clienti a cui l’azienda intende rivolgersi.
Il documento ANVUR sui parametri per l’abilitazione scientifica Giuseppe De Nicolao
Analisi critica del documento ANVUR "Criteri e parametri di valutazione dei candidati e dei commissari dell’abilitazione scientifica nazionale". Intervento alla Tavola Rotonda "Valutazione e Valorizzazione della Ricerca e dei Ricercatori" - Pisa 9/9/2011 nell'ambito del Convegno Automatica.it 2011.
http://www.convegnoautomaticaitaliana.org/Dibattiti.htm
4. Information Retrieval vs
Information Filtering
IR
IF
Representation
of information
needs
Queries
User profiles
Goal
Selecting relevant items
(docs) that match a
query
Filtering out the many
irrelevant data items
in accord with a user's
profile
Type of use
Ad-hoc use
Repetitive use
Type of users
One-time users
Long-term users
Index
Items
User profiles
Database
Relatively static
Dynamic
U. Hanani, B. Shapira, P. Shoval. “Information Filtering: Overview of Issues, Research and Systems”. User
Modeling and User-Adapted Interaction, 11(3): 203-259, 2001
7. Recommender Systems
Information filtering personalizzato
selezione di item fra una miriade di possibilità, in
base a interessi e necessità degli utenti
suggeriscono interazioni con nuovi item analizzando
le passate interazioni
9. Recommender Systems
La progettazione richiede conoscenze di
varie discipline
statistics, machine learning,
human-computer interaction,
social network analysis,
psychology
10. Recommender Systems
La progettazione richiede conoscenze di
varie discipline
statistics, machine learning,
human-computer interaction,
social network analysis,
psychology
14. 14
Output
Suggerisce item apprezzati da altri utenti che hanno
preferenze simili
Generalmente
una
lista di Top-N
suggerimenti: N
item considerati più
accurati
http://www.youtube.com/feed/recommended
19. 19
Content based filtering
Suggerisce item apprezzati da altri utenti che hanno
preferenze simili
Punti di forza
indipendenza dell'utente
indipendenza dal numero utenti e dalla popolarità degli item
trasparenza (è possibile fornire spiegazioni)
Limiti
sensibilità a informazioni superficiali o incomplete
over-specialization
cold-start
20. 20
Content representation
Se il contenuto è rappresentato da una
descrizione testuale, è necessaria una
strutturazione tramite tecniche NLP
Tokenizzazione
Eliminazione Stop Words
Stemming
Assegnazione di un peso ai token (tf-idf)
22. 22
Collaborative filtering
Suggerisce item apprezzati da altri forza che hanno
Punti di utenti
preferenze simili
suggerimenti diversificati per categorie di item
indipendenza dal contenuto (che può non esistere)
molto accurati secondo valutazioni empiriche
Limiti
Dipendenza dal numero di utenti
Cold-start per nuovi item e utenti
Sparsità matrice user-item
24. 24
User-based Collaborative filtering
1- Similarità calcolata fra utenti
2 – Stima rating considerando le similarità
B. Sarwar, G. Karypis, J. Konstan, J. Riedl , “ItemBased Collaborative Filtering Recommendation
Algorithms”, Proceedings of the 10th international
conference on World Wide Web, pp. 285-295, 2001.
25. 25
Item-based Collaborative filtering
1- Similarità calcolata fra item
2 – Stima rating considerando le similarità
B. Sarwar, G. Karypis, J. Konstan, J. Riedl , “ItemBased Collaborative Filtering Recommendation
Algorithms”, Proceedings of the 10th international
conference on World Wide Web, pp. 285-295, 2001.
26. 26
Altre categorie di RS
Social
Context-aware
Personality-based
Knowledge-based
Geographic
28. 28
Valutare le perfomance
Sperimentazione in vitro
Sperimentazioni con utenti
Solitamente si susseguono: molti algoritmi sono
confrontati e ottimizzati in vitro, i migliori sono
valutati con utenti
29. 29
Sperimentazione in vitro
1- scegliere un dataset
(Es. Movieles)
2 - Partizionare i rating di ogni utente
(Es. Hold-out, Cross Validation)
3 – Per ogni (o qualche) utente nel dataset il
RS è addestrato sull'intero dataset esclusi i
rating dell'utente considerato
4 – I suggerimenti del RS sono confrontati con
i rating di test dell'utente
30. 30
Valutare l'accuratezza
Indica il grado di corrispondenza dei suggerimenti
ad interessi e necessità degli utenti
Metriche di errore
Metriche di classificazione
33. 33
Diversity
Individual Div
Aggregate Div
Definition
diversity of
recommendation sets
for a given
individual user
diversity of
recommendations across
all users
Resolve
Over-specialization
problem
Rich-get-richer
phenomenon
Benefit
User-experience
Sales
35. 35
G. Adomavicius, Y. Kwon , “Improving Aggregate Recommendation Diversity
Using Ranking-Based Techniques”, IEEE Transactions on Knowledge and Data
Engineering, vol. 24. no. 5, pp. 896 - 911, 2012
38. 38
M. Drosou and E. Pitoura, "Comparing diversity heuristics", Technical Report, Computer
Science Department, University of Ioannina, 2009
39. 39
MMR – Maximal Marginal Relevance
Considera sia l'accuratezza che la distanza.
È piuttosto efficiente ed efficace.
Un limite è l'assunzione di indipendenza fra
rilevanza e diversità
41. 41
Aggregate Diversification
Euristiche di re-ranking
G. Adomavicius, Y. Kwon , “Improving Aggregate Recommendation Diversity Using
Ranking-Based Techniques”, IEEE Transactions on Knowledge and Data Engineering, vol.
24. no. 5, pp. 896 - 911, 2012
Euristiche basate sulla teoria dei grafi
G. Adomavicius, Y. Kwon , “Maximizing Aggregate Recommendation Diversity: A
Graph-Theoretic Approach ”, Proceedings of Workshop on Novelty and Diversity in
Recommender Systems, Chicago, Illinois, USA, pp. 3-10, 2011