1 / 21

Corpus redundancy and bias in statistical NLP

Corpus redundancy and bias in statistical NLP. Raphael Cohen, Michael Elhadad Noemie Elhadad. What is Natural Language Processing?. If it has to do with human readable (more or less) text – it NLP! Search engines. Information extraction. Helping the government read your emails.

mmunsell
Download Presentation

Corpus redundancy and bias in statistical NLP

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Corpus redundancy and bias in statistical NLP Raphael Cohen, Michael Elhadad Noemie Elhadad

  2. What is Natural Language Processing? • If it has to do with human readable (more or less) text – it NLP! • Search engines. • Information extraction. • Helping the government read your emails. • Topic Models. • Movie reviews aggregators. • Spell chekers. • …

  3. Algorithms based on occurences • Detecting collocations: "קפה עלית", “כאב ראש“Dunning 1994 – Word occurrences, Chi-Square / Maximum Likelyhood • Topic Modeling: “לידה / הריון“ vs "טפיל"Blei et al. 2003 – Mixed generative model acquired using Gibbs sampling over word occurrences in document.

  4. Electronic Health Records - Background • Hospital data is becoming digital. • Textual part of EHR is important. In our Hebrew collection of 900 neurology notes – only 12 prescriptions are indexed. • This data is used for a variety of purposes: Discovering drug side effects (Saadon and Shahar), discovering adverse drug relations, creating summaries for physicians in hospitals, studying diseases and more.

  5. Redundancy in Electronic Health Records • Observation:Physicians like to copy/paste previous visits to save time (couldn’t do it with paper notes). • Wrenn et al. showed up to 74% redundancy. It occurs in the same patient notes (Thank god…), usually within the same form but not always.

  6. Patient notes? Is this a bioinformatics talk? • No fear, other interesting datasets are also redundant:News reports (try Google News)Movie reviewsProduct reviewsTalkbacks in Ynet… • Also, we call ourselves Medical-Informatics, and have our own conferences.

  7. Do we have redundancy in our dataset? On average 52% identity, but we can see two document populations.

  8. Is redundancy a problem? • Conventional wisdom – the more data the better performance of statistical algorithms. • This usually works for huge corpora (the internet). • To solve domain specific problems we have to use smaller corpora (For example, translating CS literature from English to Chinese) • However, redundancy creates false occurrence counts. With some patients having hundreds of redundant notes, this might create a bias in smaller corpora.

  9. Our Corpus • 22,564 patient notes of patients with kidney problems. • 6,131,879 tokens. • The physician tells us that the most important notes are those from the “primary-health-care-provider” table in the database. • There are 504 patients with such notes, and 1,618 “primary-provider” notes.

  10. Let’s see it with Graphs! Effect on word counts

  11. Annotating the text • Medical concepts are detected using Health-Term-Finder, an NLP program based on the OpenNLP suite and UMLS (Unified Medical Language System) a medical concept repository. • These concepts include drugs, findings, symptoms… • Hey, you said no bio… - annotations are used with names of actors (movie reviews / gossip), corporations (news) and terrorists (online forums and chats).

  12. Graphs! You are becoming very very convinced Effect on UMLS concept counts

  13. Graphs! Effect on co-occurrence in UMLS concepts

  14. What can we do? • Build a corpus with controlled amount of redundancy. • Reminiscent of Non-Redundant protein/DNA databases built in the beginning of the last decade [Holes and Sanders (1998)].

  15. How do I get me one of them non-redundant databases? • Our easy and naïve approach:We have the patients’ ids. Let’s sample a small number of notes from each patient (The “Last” dataset in the graphs we saw). • Drawbacks:a) Annonimized data-sets are the future (our Soroka collection is on example)- they ain’t got ids.b) Are we throwing out some good data along with the redundant stuff?

  16. What did they do in Bioinformatics*? • Align all pairs of sequences (Nimrod showed us how to do that last week) and kick out the redundant ones. • Problem: Alignment costs ~O(n²), this will take a while. • Solution: BLAST / FASTA algorithms use short identical finger prints (substrings) to only compare sequences likely to be similar and to cut down O(n²) to ~O(n) in most cases. *Experts say that using borrowed algorithm from another discipline gets you into journals

  17. What’s the difference? • The Bioinfo algorithms are optimized for 4/20 (now 21) alphabets, and the sequences are shorter (usually less than 5K characters). • Texts are easier than DNA, the have defined end of lines and only one reading frame. • Fingerprinting methods for texts already exist in order to find plagiarism.

  18. a gReedy Algorithm for Controlled redundancy cOrporacoNstruction (RACOON) Sort documents by size. For each document: Find finger prints by lines (For each line, break into substrings of length F) Add to the corpus if there is no document sharing more than Max_redundancy substrings in the corpus

  19. We don’t align, is it any good?

  20. Fast facts • How long does it take?5 minutes for our 20K documents.20 minutes for our 400k documents. • Is it better than the “Last note” naïve approach?

  21. Thank You!

More Related