7. March 2003Leif Grönqvist, MSI - Växjö1 Latent Semantic Indexing and Beyond Leif Grönqvist School of Mathematics and Systems Engineering.

Slides:



Advertisements
Liknande presentationer
1.Numerical differentiation and quadrature Discrete differentiation and integration Ordinary.
Advertisements

Vägledningscentrum Career guidance centre
För att uppdatera sidfotstexten, gå till menyfliken: Infoga | Sidhuvud och sidfot Fondbolagsträff 2015.
Hjälper huvudverbet att ange tid eller modalitet
Public Participation and Dialogue in Road-Planning and Road-Design Example: Bypass Norrtälje Suzanne de Laval SWEDEN.
Arkitektrollen. Ansvar och uppgifter Architecture notebook Mycket intensivt elaboration – inception Mål: en stabil arkitektur i slutet på elaboration.
Friday 30. May 2003NoDaLiDa 2003: Leif Grönqvist1 Latent Semantic Indexing and Beyond Leif Grönqvist School of Mathematics and Systems.
Stockholm: Dec -03Vector-based Semantic Analysis1 Vector-based Semantic Analysis 1 Leif Grönqvist Växjö University (Mathematics and.
 Who frågar efter en persons (eller personers) identitet (vem dem är).  Who is he?  Who are they?  Who is coming?
To practise speaking English for 3-4 minutes Genom undervisningen i ämnet engelska ska eleverna ges förutsättningar att utveckla sin förmåga att: formulera.
© Gunnar Wettergren1 IV1021 Project models Gunnar Wettergren
1-1 Copyright © 2009 Pearson Education, Inc. Publishing as Pearson Addison-Wesley 1-1 Programmering 7.5 hp Programmering är... creativ, fascinerande, roligt,
INLÄGG FÖR SOCIALA MEDIER SHAREWEAR. 1.INSTAGRAM 2.FACEBOOK 3.TWITTER INNEHÅLL.
Don´t just try! Do! Emma Nääs
Lab Contact 1  Lab Assistants:  Meng Liu, Group B  Sara Abbaspour, Group A
Skolkör år 4-9 Skolkören övar inför avlsutningen.
Gränsöverskridande samarbete över fjäll och hav Min ansökan.
THINGS TO CONSIDER WHILE PLANNING A PARTY Planning an event can take an immense amount of time and planning. Even then, the biggest problem that arises.
STEPS TO FOLLOW FOR BECOMING A SHIP CAPTAIN A career as a ship captain can be a tedious task. Ship captains take care of business, navigation and operation.
SAFETY EQUIPMENT USED IN MARITIMEOPERATIONS One of the most important sections in maritime courses consists of boat and ship operations. Safety is an important.
Advice from Bronx Best Real Estate Attorney. Jagiani Law office of New York has been successfully working as divorce attorney & Real estate attorney for.
Digitization and Management Consulting
Why you should consider hiring a real estate attorney!
Law abiding grounds of filing a divorce Jagianilaw.com.
Types of Business Consulting Services Cornerstoneorg.com.
Tennis as they see it Research on attitudes to tennis of junior tennis players through gender perspective.
Informationssäkerhet
Mathematics 1 /Matematik 1
Nationella provets delar:
Bringapillow.com. Online Dating- A great way to find your love! The words ‘Love’ and ‘Relationship’ are close to every heart. Indeed, they are beautiful!
Alternativ till DiVA? Elisabeth Uhlemann,
Work of a Family law attorney Jagianilaw.com. A Family Law Attorney basically covers a wide range spectrum of issues that a family may face with difficulty.
The Online Way to Engagement and Wedding Jewelry! Pearlleady.com.
Meeting singles had never been so easy before. The growing dating sites for singles have given a totally new approach to getting into relationships. ‘Singles.
Hoppas det här går hem ! Bildspelet vecka 3 5 BE ® BrucElvis
We are the world There comes a time When we head a certain call When the world must come together as one There are people dying And it's time to lend.
Waste management on export
Strategic Sustainable Development
Formal Languages, Automata and Models of Computation
My role model.
Role of Divorce, Family Law and Commercial Attorneys.
Svarsfrekvensen i undersökningar från webbpaneler. Några resultat
Pearlleady.com Attractive Graduation and Wedding Gifts Online.
How to Buy Engagement Rings for Women Online?. Buying engagement rings for women or tiffany celebration rings from the online market could be a bit challenging.
Amazing Wedding/Bridal Jewellery & Gifts Available Online Pearlleady.com.
A. Speaking B.1. Reading B.2. Listening C. Writing.
You Must Take Marriage Advice to Stop Divorce! Dontgetdivorced.com.
Nationella provets delar:
Practice and challenges in involving fathers
Season 2018.
Accounts + SD = ♥? SD indicators generated from an integrated statistical account New report financed by Eurostat, DG Environment and Statistics Sweden.
Eunis Research and Analysis Initiative
Publish your presentations online we present SLIDEPLAYER.SI.
Publish your presentations online we present SLIDEPLAYER.RS.
Publish your presentations online we present SLIDEPLAYER.IN.
Publish your presentations online we present SLIDEPLAYER.VN.
Publish your presentations online we present SLIDEPLAYER.RO.
Publish your presentations online we present SLIDEPLAYER.EE.
Publish your presentations online we present SLIDEPLAYER.CO.IL.
Publish your presentations online we present SLIDEPLAYER.AE.
Publish your presentations online we present SLIDEPLAYER.BG.
Publish your presentations online we present SLIDEPLAYER.AFRICA.
Publish your presentations online we present SLIDEPLAYER.MX.
Publish your presentations online we present SLIDEPLAYER.LT.
Publish your presentations online we present SLIDEPLAYER.LV.
Publish your presentations online we present SLIDEPLAYER.SK.
Packaging that makes life easier!
”Frust” Presenteres
Presentationens avskrift:

7. March 2003Leif Grönqvist, MSI - Växjö1 Latent Semantic Indexing and Beyond Leif Grönqvist School of Mathematics and Systems Engineering the Swedish Graduate School of Language Technology

7. March 2003Leif Grönqvist, MSI - Växjö2 Overview My background Introduction to vector space models and Latent Semantic Indexing A toy example Interpretation Some applications A concrete example and a small experiment Improvements of the model Various unsolved problems Conclusion: things I have to do

7. March 2003Leif Grönqvist, MSI - Växjö3 My Background : ”4-årig teknisk” (electrical engineering) : MSc (official translation of “Filosofie Magister”) in Computing Science, Göteborg University : 62 points in mechanics, electronics, etc : Work at the Linguistic department in Göteborg Various projects related to corpus linguistics Some teaching on statistical methods (Göteborg and Uppsala), and corpus linguistics in Göteborg, Sofia, and Beijing 1995: Consultant at Redwood Research, in Sollentuna, working on information retrieval in medical databases : Work at the department of Informatics in Göteborg (the Internet Project) : PhD Student in Computer Science / Language Technology

7. March 2003Leif Grönqvist, MSI - Växjö4 Vector Space Models If we had a way to map any term to a vector in a high-dimensional space, in a way such that the similarity between the meaning of the terms is reflected in the distance between the vectors… Then we could: For a given term t, find an ordered list of the terms most similar to t For any two terms, find the similarity between them

7. March 2003Leif Grönqvist, MSI - Växjö5 Vector Space Models, cont. And if it is possible to add meaning for terms and this is also reflected by adding the corresponding vectors, we could do some more things: If we assume that it is possible to extract terms from a document, we can map documents to vectors too! A set of terms (one or more terms) may be seen as a document as well

7. March 2003Leif Grönqvist, MSI - Växjö6 Vector Space Models, cont. Now it is possible for any [term or document] d, to find an ordered list of the terms or documents most similar to d Further, we can for any two [term or document]s, find the similarity between them Therefore it is meaningful to look at terms as a special case of document – a short one

7. March 2003Leif Grönqvist, MSI - Växjö7 Alternative data sources A useful data source to get similar information would be a thesaurus, a WorldNet, or any kind of knowledge database. But: We don’t have them for all languages They are not domain specific and domain specific terms are not covered In such data sources most of the words are missing Especially names, compounds, technical terms and numbers My big newspaper corpus contains ~ unique words A vector space model can be trained from raw un- annotated corpus data!

7. March 2003Leif Grönqvist, MSI - Växjö8 Calculating a vector space The training process needs a large set of documents - the bigger the better. My data set used for experiments contains roughly 1.5 million newspaper articles and 0.5 billion running words but I will collect more… Step 1: Create a word-by-document matrix - each element in the matrix is a frequency for a word type in a specific document From here there are several ways to find a good vector space

7. March 2003Leif Grönqvist, MSI - Växjö9 Vector Space Algorithms Singular Value Decomposition (SVD) This is a mathematically complicated (based on eigen-values) way to find an optimal vector space in a specific number of dimensions Computationally heavy - maybe 20 hours for my test set Uses often the entire document as context Random Indexing (RI) Select a number of dimensions randomly Not as heavy to calculate, but more unclear (for me) why it works Uses a small context, typically 1+1 – 5+5 words Neural nets, Hyperspace Analogue to Language, etc.

7. March 2003Leif Grönqvist, MSI - Växjö10 The terminology I use Some people use these terms in a sloppy way. For me: LSI=LSA: Latent Semantic Indexing/ Analysis are used in roughly the same way by most people Two ways to obtain the model used in LSA are SVD and RI – they both find the latent information

7. March 2003Leif Grönqvist, MSI - Växjö11 The distance measure Three easy-to-calculate distance measures: Cosine: the cosine of the angle between the vectors Euclidean distance: just the distance as we all know it Manhattan distance: the distance if you walk only along the orthogonal axes Just as easy to calculate in n dimensions where n>>3 The most used is the cosine

7. March 2003Leif Grönqvist, MSI - Växjö12 A toy example

7. March 2003Leif Grönqvist, MSI - Växjö13 What SVD gives us X=T 0 S 0 D 0 : X, T 0,S 0,D 0 are matrices

7. March 2003Leif Grönqvist, MSI - Växjö14 And our example: T

7. March 2003Leif Grönqvist, MSI - Växjö15 And our example: S

7. March 2003Leif Grönqvist, MSI - Växjö16 And our example: D

7. March 2003Leif Grönqvist, MSI - Växjö17 We can recalculate X with m=2 C1C2C3C4C5M1M2M3M4 Human Interface Computer User System Response Time EPS Survey Trees Graph Minors

7. March 2003Leif Grönqvist, MSI - Växjö18 What does the SVD give? Dumais 1995: “The SVD program takes the ltc transformed term-document matrix as input, and calculates the best "reduced-dimension" approximation to this matrix.” Michael W Berry 1992: “This important result indicates that A k is the best k-rank approximation (in at least squares sense) to the matrix A. Leif: What Berry says is that SVD gives the best projection from n to k dimensions, that is the projection that keep distances in the best possible way, so no problems with local maxima.

7. March 2003Leif Grönqvist, MSI - Växjö19 What does it really mean then? The fact that a word w is represented by a specific vector v means exactly nothing! If two words a, b are represented by vectors close to each other (the angle between them is small) then: a and b are often found in the same document and/or a is often found together with c and c is often found together with b And so on…

7. March 2003Leif Grönqvist, MSI - Växjö20 A naïve algorithm Not trivial that SVD and RI works. I will explain a naive but more intuitive algorithm to obtain a result similar to SVD, but too slow for practical use: 1. Select a random point in a space with the selected dimensionality, for each unique word 2. For each document D in the set: move the points corresponding to each word towards the mass center for the words/points in D. 3. If any point made a “big” move since last iteration, then go back to step 2 Step 1-3 could be done several times to have a chance to find the global maximum

7. March 2003Leif Grönqvist, MSI - Växjö21

7. March 2003Leif Grönqvist, MSI - Växjö22 Zoom into the blue cluster

7. March 2003Leif Grönqvist, MSI - Växjö23 And the red one

7. March 2003Leif Grönqvist, MSI - Växjö24 Some applications Automatic generation of a domain specific thesaurus Keyword extraction from documents Find sets of similar documents in a collection Find documents related to a given document or a set of terms

7. March 2003Leif Grönqvist, MSI - Växjö25 Problems and questions How can we interpret the similarities as different kinds of relations? How can we include document structure and phrases in the model? Terms are not really terms, but just words Ambiguous terms pollute the vector space How could we find the optimal number of dimensions for the vector space?

7. March 2003Leif Grönqvist, MSI - Växjö26 An example based on 5000 newspaper articles pelle svensson pelle svensson svenssons ödsligt skandal frikännande polismannens tjänstetid slutkörd munsex avstyra bengt johansson johansson bengt davidson folkpartiledaren kdsledaren ö resundsbroprojektet centerledaren irhammar partiledarna avgaser lyckosamt

7. March 2003Leif Grönqvist, MSI - Växjö27 Bengt Johansson is just Bengt + Johansson – something is missing! bengt folkpartiledaren westerberg kdsledaren riksdagsledamot ändrats ingbritt irhammar tolkningen tolkar partiledarna johansson olof miljödepartementets görel thurdin miljöminister brofrågan rosenbad miljödepartementet regeringssammanträdet avgaser

7. March 2003Leif Grönqvist, MSI - Växjö28 A small experiment I want the model to know the difference between Bengt and Bengt 1. Make a frequency list for all n-tuples up to n=5 with a frequency>1 2. Keep all words in the bags, but add the tuples, with space replaced by _, as words 3. Run the LSI again Now Bengt_Johansson is a word, and Bengt_Johansson is NOT Bengt + Johansson Number of terms grows from to

7. March 2003Leif Grönqvist, MSI - Växjö29 New results Some distances bengt_johanssonjohansson bengt_johansson bengt bengt_johanssonolof bengt_johanssonfolkpartiledaren johanssonolof johanssonfolkpartiledaren johanssonbengt bengtfolkpartiledaren bengtolof folkpartiledarenolof

7. March 2003Leif Grönqvist, MSI - Växjö30 And the top list for Bengt_Johansson bengt_johansson handbollslandslag gunnar_blombäck fyrnationsturneringen_i_östergötland fyrnationsturneringen förbundskapten_bengt_johansson förbundskapten_bengt blombäck carlen åtta_mål bänken magnus_wislander wislander målet_stod svenske_förbundskaptenen orutinerade vinna_den_här magnus_andersson matchen_spelades förbundskaptenen landskamp glädjeämnen vmlaget halvlek världsstjärnor bottenlaget brolin uppvisningen offensivt jörgensen landslag

7. March 2003Leif Grönqvist, MSI - Växjö31 The new vector space model It is clear that it is now possible to find terms closely related to Bengt Johansson – the handball coach But is the model better for single words or for document comparison? What do you think? More “words” than before – hopefully it improves the result just as more data does At least no reason for a worse result... Or?

7. March 2003Leif Grönqvist, MSI - Växjö32 An example document REGERINGSKRIS ELLER INTE PARTILEDARNA I SISTAMINUTEN ÖVERLÄGGNINGAR OM BRON Under onsdagskvällen satt partiledarna i regeringen i sista minutenöverläggningar om Öresundsbron Centerledaren Olof Johansson var den förste som lämnade överläggningarna På torsdagen ska regeringen ge ett besked Det måste dock enligt statsminister Carl Bildt inte innebära ett ja eller ett nej till bron …

7. March 2003Leif Grönqvist, MSI - Växjö33 Closest terms in each model 0.986underkänner 0.982irhammar 0.977partiledarna 0.970godkände 0.962delade_meningar 0.960regeringssammanträde 0.957riksdagsledamot 0.957bengt_westerberg 0.954materialet 0.952diskuterade 0.950folkpartiledaren 0.949medierna 0.947motsättningarna 0.946vilar socialminister_bengt_westerberg 0.967partiledarna 0.921miljökrav 0.921underkänner 0.918tolkar 0.897meningar 0.888centerledaren 0.886regeringssammanträde 0.880slottet 0.880rosenbad 0.877planminister 0.866folkpartiledaren 0.855thurdin 0.845brokonsortiet 0.839görel 0.826irhammar

7. March 2003Leif Grönqvist, MSI - Växjö34 Closest document in both models BILDT LOVAR BESKED OCH REGERINGSKRIS HOTAR Det blir ett besked under torsdagen men det måste inte innebära ett ja eller nej från regeringen till Öresundsbroprojektet Detta löfte framförde statsminister Carl Bildt under onsdagen i ett antal varianter Samtidigt skärptes tonen mellan honom och miljöminister Olof Johansson och stämningen tydde på annalkande regeringskris De båda har under den långa broprocessen undvikit att uttala sig kritiskt om varandra och därmed trappa upp motsättningarna Men nu menar Bildt att centern lämnar sned information utåt Johansson och planminister Görel Thurdin anser å andra sidan att regeringen bara kan säga nej till bron om man tar riktig hänsyn till underlaget för miljöprövningen …

7. March 2003Leif Grönqvist, MSI - Växjö35 DocBasic modelPhrases added ScoreRankScoreRank

7. March 2003Leif Grönqvist, MSI - Växjö36 Documents with better ranking in the basic model BRON KAN BLI VALFRÅGA SÄGER JOHANSSON Om det lutar åt ett ja i regeringen av politiska skäl då är naturligtvis den här frågan en viktig valfråga … INTE EN KRITISK RÖST BLAND CENTERPARTISTERNA TILL BROBESKEDET En etappseger för miljön och centern En eloge till Olof Johansson Görel Thurdin och Carl Bildt …

7. March 2003Leif Grönqvist, MSI - Växjö37 Documents with better ranking in the phrase model ALF SVENSSON TOPPNAMN I STOCKHOLM Kds-ledaren Alf Svensson toppar kds riksdagslista för Stockholms stad och Michael Stjernström sakkunnig i statsrådsberedningen har en valbar andra plats … BENGT WESTERBERG BARNPORREN MÅSTE STOPPAS Folkpartiledaren Bengt Westerberg lovade på onsdagen att regeringen ska göra allt för att stoppa barnporren …

7. March 2003Leif Grönqvist, MSI - Växjö38 Hmm, adding n-grams was maybe too simple... If the bad result is due to overtraining, it could help to remove the words I build phrases of, but maybe not all Another way to try is to use a dependency parser to find more meaningful phrases, not just n-grams

7. March 2003Leif Grönqvist, MSI - Växjö39 The interpretation of similarities I havn’t tried to solve this problem at all but one idea I have is to: Calculate vector spaces for various dimensionalities and context widths Check if the different settings find different kind of relations With a data source like WordNet it could be done in a systematic way

7. March 2003Leif Grönqvist, MSI - Växjö40 How to select the number of dimensions Susan T Dumais 1995: “In previous experiments we found that performance, improves as the number of dimensions is increased up to 200 or 300 dimensions, and decreases slowly after that to the level observed for the standard vector EC­3 method (Dumais, 1991).” Jason I Hong 2000: “There does not seem to be a general consensus for an optimal number of dimensions; instead, the size of the concept space must be determined based on the specific collection of documents used.” Thomas K Landauer 1997: “Near maximum performance of 45-53%, corrected for guessing, was obtained over a fairly broad region around 300 dimensions” Leif 2003: “We should try to do similar experiments as Dumais/Landauer, but relate the optimal dimensionality to measures like number of documents, terms, nonzero elements, etc, because these could give us a formula not relying on hand tagged data sets”

7. March 2003Leif Grönqvist, MSI - Växjö41 Performance for the SVD Dumais 1995: “The SVD takes only about 2 minutes on a Sparc10 for a 2k x 5k matrix, but this time increases to about hours for a 60k x 80k matrix.” Hong 2000: “The SVD algorithm is O(N 2 k 3 ), where N is the number of terms plus documents, and k is the number of dimensions in the concept space”, “However, if the collection is stable, SVD will only need to be performed once, which may be an acceptable cost.” Leif: So if a good computer today is 100 times faster than Dumais’ 1995 and we have 20 times bigger data sets and we have an optimized SVD function instead of a research prototype, it should still take around 20 hours.

7. March 2003Leif Grönqvist, MSI - Växjö42 What I still have to do something about Find a better LSI/SVD package than the one I have (old C-code from 1990), or maybe writing it myself... Get the phrases into the model in some way When these things are done I could: Try to interpret various relations from similarities in a vector space mode Try to solve the “number of optimal dimensions”- problem Explore what the length of the vectors mean