Marking time in developmental biology

In developmental biology, to support reasoning about cause and effect, it is critical to link genetic pathways with processes at the cellular and tissue level that take place beforehand, simultaneously or subsequently. While researchers have worked on resolving with respect to absolute time, events mentioned in medical texts such as clinical narratives (e.g. Zhou et al, 2006), events in developmental biology are primarily resolved relative to other events. In this regard, I am developing a system to extract and time-stamp event sentences in articles on developmental biology, looking beyond the sentence that describes the event and considering ranges of times rather than just single timestamps. I started by creating four gold standard corpora for documents, event sentences, entities and timestamped events (for future public release). These datasets are being used to develop an automated pipeline to (1) retrieve relevant documents; (2) identify sentences within the documents that describe developmental events; and (3) associate these events with the developmental stage(s) that the article links them with or they are known to be linked with through prior knowledge. Different types of evidence are used in each step. For determining the relevant developmental stage(s), the text surrounding an event-containing sentence is an efficient source of temporal grounding due of its immediate accessibility. However, this does not always yield the correct stage and other sources need to be used. Information within the sentence, such as the entities under discussion, can also be used to help with temporal grounding using mined background knowledge about the period of existence of an entity.