February 9
13.00-17.00
Röd

17.00-18.00

Orange

Lecture 4: IR Beyond one shot

Jussi Karlgren

Manning Chapter 9, Robertson

Read after Lecture 5

Lecture 5: Scoring, weighting, vector space model
Hedvig Kjellström

Manning Chapter 6, 7

Read before Lecture 4

Lecture 8: Some useful additions to a search engine, Random Indexing
Viggo Kann

April 5
13.00-18.00

Orange

Lecture 9: Guest lecture

Hercules Dalianis, SU

Clinical text retrieval - some methods and some applications

Electronic patient records contain a waste source of information, both in form of structured information as diagnosis codes, drug codes, lab values, time stamps, etc and unstructured in form of free text. Methods - both rule based and machine learning based for retrieving this information is presented. Applications as diagnosis codes assignment, hospital acquired infection detection and adverse drug event detection will be discussed.

Lecture 10: Guest lectures

Maria Skeppstedt, Linnaeus University and Gavagai

Sentiment, stance and applications of distributional semantics

Sentiment analysis aims at determining the attitude of the speaker/writer, (e.g., 'positive' or 'negative'), while stance detection aims at determining the speaker's/writer's position on an issue (e.g., 'for' or 'against' an issue). What methods could you use for sentiment analysis/stance detection and what are the practical applications? How could you use distributional semantics? Examples from products from the Swedish language technology company Gavagai will be given, as well as from the StaViCTA project at Linnaeus University.

Roelof Pieters, KTH and Graph Technologies

Multi-modal Retrieval and Generation with Deep Distributed Models

Multi-modal retrieval is emerging as a new search paradigm that enables information retrieval from various types of media. For example, users can simply make a photo of a movie poster to search for relevant reviews and trailers. Key to such approaches are very deep neural networks which can be trained to represent such diverse data extremely well. Once "embedded" in such a space, one can even turn such neural networks upside down to create new novel data. Imagine a neural network that can compose songs, or write novels in the style of Shakespeare!
The talk will give a short overview of Deep Learning approaches, multi-modal embedding methods, to then show some examples of recent and ongoing work in multimodal search and generation.

April 29

9.00-12.00

Brun

Extra computer hall session

Lecture 11: Guest lectures

Boxun Zhang, Spotify

Introduction to Locality Sensitive Hashing

In this lecture, I will first give a brief introduction of locality sensitive hashing,
a widely-used algorithm for nearest neighbor search, hierarchical clustering,
audio fingerprint, and etc. Then, I will describe how we use this algorithm at
Spotify.

Anders Friberg, KTH

Music Information Retrieval

The recent paradigm shift in music distribution and production has created a need for developing methods of music audio content analysis. This is one of the major directions within the field music information retrieval and its aim is to understand music audio similar to how human listeners perceive music. An overview of methods and challenges in the field with some snapshots from KTH research will be presented.

Lecture 12: Guest lectures

Simon Stenström, Findwise

Search From the Trenches

What distinguished a search engine from a search solution? This talk focuses on the part of findability that a search engine just doesn't do well on its own. How can we solve that? What else do you need?

Magnus Rosell, FOI

Classified