Experiences on integrating explicit knowledge on information access tools in the medical domain

+ Computer- Query user-
aided defined
summarization expansion

Post-retrieval Extractive
clustering Summarization

Experiences on integrating explicit knowledge on
information access tools in the medical domain

Manuel de la Villa
Department of Information Technologies
University of Huelva

+ 2

Index

 Brief CV
 Why a research stay? In Wolverhampton?
 Teaching

 Integrating explicit knowledge on information
access tools
 Knowledge sources (UMLS & Freebase)
 Automatic Text Summarization
 Information Retrieval

Research Group in Computational Linguistics (Univ. Wolverhampton), June 20th 2011

+ 3

Brief CV


+ 6

Teaching experience

 Software Engineering
 Process and Methodologies, Metrics,
Requirements analysis, Design, …
 Software Engineering Lab (UML, NetBeans,
Subversion, Java, JUnit, Persistence…)

 Multimedia applications development
 Adobe Director, Flash, Photoshop, Premiere
 Sony Sound Forge, Audacity


+ 7

Knowledge integration


+ Specific Domain Knowledge source. UMLS (I) 8

ICD-10
LOINC

SNOMED-CT UK-Clinical Terms
UMLS MeSH
DSM-IV
…
Gene Ontology RxNorm

An homogeneus group of terminologies A saturation of different terminologies

UMLS aims to overcome a significant barrier, the variety of
ways the same concepts are expressed in different
machine-readable sources.

+ Specific Domain Knowledge source. UMLS (II) 9

Project NLM Unified Medical Language System (UMLS):

 Aim, to develop tools that help researchers in the knowledge
representation, retrieval and integration of biomedical information.
 UMLS Knowledge Sources ‫‏‬

 Software tools
Three main components:

SPECIALIST Lexicon: Compilation of lexical elements (>200.000) with grammatical
information and linguistic variants.

“Anaesthetic” “Anaesthetic”
{base=anesthetic {base=anesthetic
spelling_variant=anaesthetic spelling_variant=anaesthetic
entry=E0330018 cat=noun entry=E0330019 cat=adj
variants=reg variants=uncount } variants=inv position=attrib(3)
position=pred stative }

+ Specific Domain Knowledge source. UMLS (III) 10

 Metathesaurus: very large, multi-purpose, and multi-lingual
vocabulary database (compiles more than 100 source
vocabularios), https://uts.nlm.nih.gov/metathesaurus.html
 every term (>5M) associated with a concept (>1.5M), terms
related (e.g., synonyms) (16M relations)

 each concept assigned to one or more semantic types of the 135
existing
Different terms…

for a same concept…

Included in a semantic type

+ Specific Domain Knowledge source. UMLS (IV) 11

https://uts.nlm.nih.gov/semanticnetwork.html

 UMLS Semantic Network: is an ontology with 135
semantic types and to 54 types of relationships
between types


+ General Domain Knowledge Source: Freebase (I)

 Freebase is a large public database that collects three kinds of
information:
 data;

 texts ; and
 media , that references…
 …entities or topics (≈ 12 million). An entity is a unique single person,
place, or thing.
 A single concept or real-world thing.
 A topic could also be called an entity, resource or element or thing, it is a
fundamental unit in Freebase.
 /common/topic
 Each topic has a Guid or globally unique ID
 http://www.freebase.com/view/en/barack_obama
 http://www.freebase.com/guid/9202a8c04000641f800000000029c277

+ General Domain Knowledge Source: Freebase (II)
 Freebase connects entities together as a graph,
 defines its data structure as a set of nodes and a set
of links that establish relationships between the
nodes.
 Most of our topics are associated with one or more types (such as
people, places, books, films, etc) and may have additional
properties like "date of birth" for a person or latitude and
longitude for a location. These types and properties and related
concepts are called Schema.

+ General Domain Knowledge Source: Freebase (III)
The Schema
Schema (the way Freebase's data is laid out) is expressed through
Types and Properties. Types are grouped together in Domains.

+ General Domain Knowledge Source: Freebase (IV)
The Schema: Medicine

+ General Domain Knowledge Source: Freebase (V)
How can we use it…

 As a reference or information source
 Create interesting Views and Visualizations and
share them with others
 Embed Freebase data in your website

 Use our API or Acre, our hosted app development
platform, to build apps that use Freebase data
 Download our Data dumps

 Use Freebase's RDF for Semantic Web applications

+ General Domain Knowledge Source: Freebase (IV)
The Freebase approach

+ MQL (Metaweb Query Language)
• http://api.freebase.com/api/service/mqlread?query={"query":{"type":"/
music/artist","name":"U2","album":[]}}
• http://api.freebase.com/api/service/mqlread?query={"query":
[{"type":"/medicine/disease", "name":null, "symptoms":
{"name":"Nausea"}}]}
• Query Editor

+ 22



+ 23

Experiences in Automatic summarization (I)

+ We develop a proposal with this main
characteristics:
 Sentences extraction
 Document representation as a graph
 Centered on biomedical concepts
 Using concept frequency to measure relevance


+ 24

Experiences in Automatic summarization (II)

+ Phase I: Graph generation
Sentences and UMLS concepts identification
+ Phase II: Similarity algorithm
Concepts overlapping between sentences
(edges) means “recommendation”
+ Phase III: Ranking algorithm
Weight associated with each edge depends on
similarity
+ Phase IV: Summary building
Top ranked sentences are selected

+ 25


similarity

+ 26


similarity

+ 27


similarity

+ 28


similarity

+ 29

Automatic Summarization. Evaluation

 Evaluation
with ROUGE (based on n-grams) against generic
summarizers
 Our method obtains good results, specially with small n-grams
de la Villa, M., Maña, M.
“Propuesta y evaluación de un método de generación de
resúmenes extractivo basado en conceptos en el ámbito
biomédico”. XXV edición del Congreso Anual de la Sociedad
Española para el Procesamiento del Lenguaje Natural 2009
(SEPLN´09) San Sebastián (Sept-2009).


+ 30



+ 31

Experiences in Computer-aided
summarization(I)
 Computer-aided
summarization combines automatic
and human summarization.
 The CAS system suggest an initial summary,
selecting relevant sentences
 The human can change the sentences selection and
edit manually the summary.
 Purpose: construction of a Gold-Standard building
assistant.
 Novelty: Considering biomedical concepts distribution
(Reeve et al., 2006)


+ 32

Experiences in Computer-aided
summarization(and II)
Experience in the design
and construction of a
Gold-Standard building
assistant (or Computer-
aided summarization)

Considering biomedical
concepts distribution
(Reeve et al., 2006)

-Client-server app
-Centralized repository
-Supports PDF, XML


+ 33



+ 34

Experiences in Information Retrieval
and Post-retrieval clustering
Experience in the design and
construction of an information
retrieval system with:
• ost-retrieval clustering,
P
• rientation to biomedical
o
documents and
• obile devices
m


Search
and
Informa.on
Retrieval

Our
implementa.on
36

Document sources: Biomed Central (web crawling in progress)
Text Processing: lowercasing, stemming, stop-words ,…

Lucene for indexing…


Search
and
Informa.on
Retrieval

Our
implementa.on
(and
II)
37

… and Lucene for searching

Clustering

Our
implementa.on
38

Weka for Clustering
The post-processing clustering is to associate, according to their
similarity, a set of documents retrieved from a query in different
subsets

38

Clustering

Why
Simple-‐K-‐Means?

Clustering algorithm:
Simple-K-Means vs Expectation Maximization

Algorithms

Simple-‐K-‐means
EM

Querys
(Documents)

Ligaments
(10)
1
2

Cancer
Skin
(25)
4
12

Cancer
(46)
5
26

Disease
(62)
8
57

Time it takes to perform the grouping in seconds

K? It depends on the number of documents retrieved.

39

Visualiza.on
on
Mobile
Devices

Our
interface

Cancer skin

40

+ 41



+
and Query user-defined expansion (I)

 Userhave problems to define their information needs in a
query string (Jansen, Spink y Koshman, 2007).
 Queries containe less than three terms (75,2%) and the majority of
queries contained one (18,5%), two (32,2%)

 Methods to improve (expand) query:
 Relevance feedback.
 Local analysis or global analysis.

 Natural Language Processing Resources.

 Experiments with users show the preferences of these to
maintain control over how the query is reformulated (Belkin
et al., 2001).

+ 43

and Query user-defined expansion (II)

 Experience on using Ontologies to assist the definition of the
search string… previosly


+
and Query user-defined expansion (II)
How does it works?
 Pre-retrieval  Construction o f the Graph

+ 45

Research: Information Retrieval
(and III)
 … or using Ontologies to build an enriched concept graph that
assist the definition of the search string

http://www.uhu.es/manuel.villa/viewmed/
de la Villa, M., Garcia, S., Maña, M.
“¿De verdad sabes lo que quieres buscar? Expansión guiada visualmente
de la cadena de búsqueda usando ontologías y grafos de conceptos”.
XXVII edición del Congreso Anual de la Sociedad Española para el
Procesamiento del Lenguaje Natural 2011 (SEPLN´11) Huelva (Sept-2011).


+ 46

Tools knowns. Expectations.

 UMLS:
 Metathesaurus, Semantic Network  Ioffer my collaboration if
 Tools: you’re interested in using
 Metamap, any of these resources
 MMTx API,
 I’mopen to collaborate on
 Semrep
whatever task you
 UTS Web Services, …
consider related and…
 Freebase
 … to receive some
 MQL (Metaweb Query Language) guidelines to improve
summarization method
 Newbie with UIMA & GATE
Any questions?

Experiences on integrating explicit knowledge on information access tools in the medical domain

More Related Content

Similar to Experiences on integrating explicit knowledge on information access tools in the medical domain (20)

More from Manuel de la Villa (15)

Experiences on integrating explicit knowledge on information access tools in the medical domain