GIS as a Narrative Generation Platform (Deep Maps), Książki
[ Pobierz całość w formacie PDF ]
//-->See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/267212470GIS as a Narrative Generation PlatformChapter · January 2015DOI: 10.13140/2.1.3821.8249CITATIONSREADS2913 authors, including:May YuanUniversity of Texas at Dallas67PUBLICATIONS1,019CITATIONSSEE PROFILEAvailable from: May YuanRetrieved on: 14 April 2016Edited Manuscript, Indiana University Press<CN>8<\><CT>GIS as a Narrative Generation Platform<\><AU>May Yuan, John McIntosh, and Grant DeLozier<\><A>Introduction<\>Maps have long been one of the key tools to represent the landscape within whichhistories occurred. While being static, maps present the spatial dimension of historical data andreveal spatial associations among spatial features of interest. Much research in spatial histories orhistorical geographical information systems (GIS) rises to the challenge of visualizing historicalsocial data, geocoding historical cultural landmarks, and analyzing their spatial patterns overtime.1Yet, historical investigations go far beyond thematic or statistical mapping. Historian JohnGaddis noted that historians exerciseselectivity, simultaneity,andshifting of scaleinmanipulation of space and time to construct narratives that interpret the past.3Selectivity isnecessary so that historians can simplify a complex reality into something manageable for astudy. When selected events expand over space and time, historians examine multiple places atonce (i.e., simultaneity) and shift scales when they use a particular episode to make a generalpoint.Scale shiftingis a fundamental tool for narration in history, andsimultaneityleads to thestudy of histories as mapping the past landscape. Landscape patterns of historical eventsconstitute the structure that historians observe in the present. Narratives are being developed ashistorians interpret the processes that produced the landscape structure. To Gaddis, historiansembed generalizations in narratives, while social scientists embed narratives in generalizations.Instead of categorical causes, historians emphasize contingent causes that are responsible fordeveloping singularities in continuity and lead to particular generalizations in history.This study on GIS as a narrative generation platform (i.e., narrative GIS) aligns well withEdited Manuscript, Indiana University PressGaddis's landscape of history. Narratives are sequential organizations of events. Whileconventional GIS centers on information characterizing geographies, a narrative GIS aims torepresent and order events in support of constructing spatial narratives. Here, spatial narrativesrefer to meaningful sequences of spatial events. Our premise is that locations where events tookplace should be considered equally important to the time of occurrences in history, and thetreatment of time should be considered as important as locations in geography. A narrative GIShenceforth aims to provide the necessary framework to make connections among events in spaceand time for narrative generation. GIS supports the need for selectivity in historical studies bydatabase queries to retrieve events of interest, and for simultaneity, by mapping selected eventsacross space and time and contextualizing these events with geographic features or other events.GIS zoom functions enable historians to shift across local, regional, and global scales insofar asthe data permit. The vision of a narrative GIS is to facilitate the interactive selection of events ofdifferent types, relate events across space and time, and assess microscopic and macroscopicstructures to embed multiple generalizations in narratives that help explain the underlyinghistorical processes.A narrative is a meaningful sequence of events, and therefore events are basic constructsfor narrative generation. Consequently, event objects are essential to a narrative GIS database.While there is no universal definition of events, we simply consider spatial event objects as aquadruple of [actor,action, location,andtime].4Actorsare entities involved in an event, andactionsevoke happenings of an event at a location and time. Actors may be biotic (e.g., humansor animals), abiotic (e.g., fires or rocks), or immaterial (e.g., communications or ideas). Actorstake actions, and actions drive changes to properties, conditions, or locations. Events are innatelytemporal; event history modeling, for example, analyzes and projects the timing and periodicityEdited Manuscript, Indiana University Pressof event occurrences.5While most events are spatial, spatial markers may not be explicit and aretreated as add-ons in most event studies.Nevertheless, many scholars recognize the importance of spatial dimensions to eventmodeling. For example, the duration of civil wars in various countries is commonly modeledwith statistical regressions on the magnitude and frequency of conflicts; yet additional spatialconsiderations led to new insights that "thegreater the frequency of states bordering the civilwar state, the longer the duration of the civil war."6In other words, the number of borderingstates has a prolonging effect on the duration of the civil war in a state. With equal treatments ofspace and time in representing and analyzing events, a narrative GIS is expected to provide newinsights into the correlation, interaction, and structure of historical events and narratives in spaceand time. Narrative generation is performed by connecting events in space and time based onactors, actions, or both to decipher the spatiotemporal relationships among actors and actions inmaking histories. As Gaddis noted, historians embed multiple generalizations in narratives.Likewise, a narrative GIS generalizes spatial events in the form of quadruples [actor, action,location, time] and embeds multiple generalizations of ordered spatial events to generatenarratives.Since the majority of historical data are documents, our development of a narrative GISbegins with ingesting text documents, extracting and assembling spatial events to form GIS eventdatabases, querying and structuring events to generate spatial narratives, and storing spatialnarratives for future queries and analyses (). As proof of concept, we use two distinctive corporaof histories in building narrative GIS databases and narrative analytics: Dyer'sCompendium ofthe War of the Rebellionand theRichmond Daily Dispatch.Frederick H. Dyers, a Civil Warveteran compiledthe Compendiumbased on materials from the Official Records of the UnionEdited Manuscript, Indiana University Pressand Confederate Armies and other sources.Figure 8.1. Workflow for narrative GIS development.Dyer's Compendium7lists organizations and movements of regiment cavalries musteredby state and federal governments for services in the Union armies. The second source ofhistorical documents is theRichmond Daily Dispatchprovided by the Digital ScholarshipLaboratory at the University of Richmond. The newspaper was one of the most widelydistributed newspapers in the south during the Civil War and included news from the entire eastcoast. TheRichmond Daily Dispatchretained the reputation of being politically unbiased andwas published throughout the Civil War. The use of the two document sources serves twopurposes. First, the writing styles are distinctive, and therefore, provide the challenge ofdeveloping algorithms of text analytics that are not specific to particular source documents.
[ Pobierz całość w formacie PDF ]