18.02.2013 Views

Timing, hosts and locations of (grouped) events of NanoImpactNet

Timing, hosts and locations of (grouped) events of NanoImpactNet

Timing, hosts and locations of (grouped) events of NanoImpactNet

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

NanoSafetyCluster - Compendium 2012<br />

6 Work performed <strong>and</strong> Results<br />

Following is a detailed description <strong>of</strong> two important components<br />

<strong>of</strong> NHECD, Information Extraction <strong>and</strong> NHECD User Interface.<br />

These components synthesize the investment <strong>of</strong> all NHECD work-<br />

packages <strong>and</strong> are reviewed in detail. The detailed description <strong>of</strong><br />

the other components was given in previous publications.<br />

6.1 Information Extraction<br />

Information Extraction (IE) is a type <strong>of</strong> information retrieval whose<br />

goal is to automatically extract structured information from<br />

unstructured <strong>and</strong>/or semi-structured machine-readable<br />

documents. In most cases, this activity concerns processing natural<br />

language texts using NLP. Information extraction has numerous<br />

potential applications. For example, information available as<br />

unstructured text can be transformed into traditional databases<br />

that users can probe through st<strong>and</strong>ard queries.<br />

6.1.1 IE system flow<br />

The IE component is built in a modular linear flow. Each module<br />

can work independently. If needed, any <strong>of</strong> the modules can be<br />

separately changed.<br />

6.1.2 The process<br />

Figure 1 -IE System Flow<br />

The aim <strong>of</strong> NHECD IE component is to extract, from every scientific<br />

paper gathered by the NHECD crawler, a comprehensive, full <strong>and</strong><br />

precise list <strong>of</strong> relations. NHECD text mining tasks are in fact<br />

information extraction tasks, namely, to extract entities <strong>and</strong><br />

relations (which are, by nature, structured information) from<br />

unstructured Nanoparticle-toxicity related documents. The<br />

information extraction system expected results include the<br />

following entities or relations: (1) Nano particle, (2) Model – Cell<br />

model or animal, (3) Attributes – NP size, Zeta potential, animal<br />

age, etc. <strong>and</strong> (4) Experiment attributes – mode <strong>of</strong> exposure,<br />

measurement assay etc.<br />

See the following example <strong>of</strong> the information extraction task.<br />

The examined text:<br />

Under phase-contrast microscope, HT-1080<br />

cells (control) appeared polyhydric or stellate<br />

showing slender lamellar expansions (Fig. 1A) that<br />

joined neighboring cells. With increasing<br />

concentration <strong>of</strong> SNP (from 6.25 to 50 µ g/mL ), cells<br />

were seen as less polyhydric; <strong>and</strong> more fusiform,<br />

shrunken <strong>and</strong> rounded.<br />

The results <strong>of</strong> XTT assays (Fig. 3) showed a dosedependent<br />

cytotoxicity for both the cell types with<br />

IC50 values <strong>of</strong> SNP working out as 10.6 <strong>and</strong> 11.6<br />

µg/mL for HT-1080 <strong>and</strong> A431 cells, respectively<br />

Figure 2- IE examined text<br />

Should result in the extraction <strong>of</strong> the following relations:<br />

Figure 3- IE relations result<br />

6.2 The NHECD User Interface<br />

The NHECD user interface <strong>of</strong>fers different possibilities to retrieve<br />

efficiently knowledge on the health, safety <strong>and</strong> environmental<br />

impact <strong>of</strong> nanoparticles.<br />

• A BASIC SEARCH interface to perform keyword <strong>and</strong>/or<br />

taxonomy based queries.<br />

• An ADVANCED SEARCH interface for experienced users to<br />

extend the capabilities <strong>of</strong> the basic search using<br />

advanced search features such as logical operators (AND,<br />

OR, NOT), as well as allowing to search through extended<br />

metadata.<br />

• The INTELLIGENT SEARCH is a unique method especially<br />

adapted to the needs <strong>of</strong> researchers in the nano-science<br />

field. This feature includes, among other capabilities, the<br />

power to search by model, experiment <strong>and</strong> nanoparticles<br />

attributes. The target is the exact data taken from the<br />

results <strong>of</strong> all experiments described in the corpus (shown<br />

in figure 4).<br />

238 Compendium <strong>of</strong> Projects in the European NanoSafety Cluster

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!