Regarding the locations of wartime events, a key point is that places are temporal instances, that is, they can change over time. Current geographical datasets such as GeoNames are thus not directly applicable to a historical case like WW1, as the documents and events often refer to the place names used at the time. Addi- tionally, for example the locations of battles often re- fer to geographical features such as ridges or rivers, as opposed to for instance villages or administrative units of the same name found more commonly in modern datasets.

..while Wikipedia contains numerous articles related to military history, many historians remain sceptical of it as a resource for either themselves or their students [19]. Indeed, while Wikipedia seems reassuringly accurate compared to traditional reference sources in the field of science [7], there is evidence to support their scepticism in the historical domain. In a field in which scholarly accountability is a core tenet, historians’ questions largely focus on Wikipedia editors’ lack of accountability, the unevenness of the qual- ity of its articles, and particularly for historical articles, a lack of references to academic sources.

To better understand the problems humanities researchers encounter in utilizing these collections, re- searchers conducted 21 semi-structured interviews with CU faculty and students [11]. The major user needs identified were better support for: 1) locating documents and data relevant to a particular topic within distributed online collections; and 2) contextualizing the content, for instance, to gauge author bias or simply become familiar with the places, events, and people mentioned in the documents. In addition, problems were identified with documents being in unfamiliar languages, as well as with ambiguities and variations in names, such as place names changing over time or a person being referenced by different names in different documents.

Historical context and uncertainty also creates complexity for temporal modeling, as it is often difficult to say with absolute precision when a certain event took place. For example, actors on opposing sides of a battle tend to see the extent of that battle differently [15], explanatory notes], depending on if they include preparatory attacks leading up to the battle or the consolidation of gains following a successful attack in the time- frame.

For this reason, CIDOC-CRM’s temporal representation supports a level of uncertainty in the encoding. In other words, this means that it is possible to present a timestamp, for instance, for "at the beginning of the year 1917" by specifying four temporal points: the earliest possible start time, latest possible start time, earliest possible end time and latest possible end time. By using such timeframes, analyses and visualizations of the temporal relationships between war events do not miss events with uncertain or ambiguous dates.