Let’s examine the various texts we saw earlier, but this time identify what percentage of their vocabulary is found in the Oxford Phrasal Academic Lexicon (OPAL). Have a look at four extracts from academic journal articles and notice the highlighted academic vocabulary from OPAL.
Again, this page is only for demonstration, but if you want you can read any of the texts and try to identify a few examples of academic vocabulary before you reveal the answer.
The second dynamic corresponds to a large number of data-driven initiatives that have emerged under the banner of open government advocacy since 2005. Web entrepreneurs, activists, civic hackers, computer enthusiasts and journalists have progressively joined forces to strengthen government accountability and citizen participation through the release of government data (Lathrop and Ruma, 2010). First in major American cities, some of them tried to promote the release of public data by municipalities. In this respect, Washington DC has been a reference with its data portal launched in 2004. Sharing the normative beliefs about freedom of information promoted by the open software movement (Holtgrewe and Werle, 2001), they have designed websites and online applications resting on city government data that give citizens the opportunity to know what is going on in their city (as regards crime, transportation, etc.) and to control their elected officials. The movement has spread to the federal level, and many data-driven online applications have been designed based on the same political concerns. Launched in 2007, http://MapLight.org, for instance, was designed by Daniel Newman – a programmer, entrepreneur and political activist – to correlate lawmakers’ voting records with the money they have received from special-interest groups (Lathrop and Ruma, 2010: 223–232).
Text adapted from Parasie & Dagiral (2013)
High-level descriptive statistics are consistent with this hypothesis; for brevity, we only discuss the proportion of bids from origin and destination states here. Because of the nature of this market, the absolute proportion of lenders is small. On average, 3.1% of bids came from lenders who were in borrowers’ origination states, and 2.8% came from lenders who were in borrowers’ destination states; both numbers are still larger than the simplistic 2% baseline in §4.1.1. More importantly, we see that before the borrowers’ moves, the proportion of bids from origination state lenders was 3.2%, and that decreased to 2.9% after the borrowers’ moves. Meanwhile, the proportion of bids from destination state lenders was 2.1% before the move, but it increased to 3.6% after (see Figure 3(a)). The pattern is highly similar if we consider the amount of bids instead: origination state lenders decreased their contributions from 3.4% to 3.0%, whereas destination state lenders increased their contributions from 2.1% to 3.7% (see Figure 3(b)).
Text adapted from Lin & Viswanathan (2016)
At the time of this study, police agencies did not have a standardized crime code, such as a Uniform Crime Reports code, to identify human trafficking offenses. As a result, we relied on study agencies to identify all investigations initiated by their agency or in which their agency cooperated regarding incidents with some evidence of human trafficking. These included cases investigated as trafficking and prosecuted either locally or federally as such, cases investigated as trafficking but prosecuted either locally or federally as a different crime, or cases investigated as trafficking but never prosecuted. A total of 254 human trafficking cases were identified across the 12 study sites. In eight of the study counties, there were fewer than 15 identified human trafficking cases. For these latter sites, we coded all available case records. In four sites, there were more than 15 identified cases. In these sites, we drew a sample of approximately 15 cases per site stratified by year and by type of trafficking to ensure representation of different case types throughout the study period. This allowed for a comparison across sites with different types of legislative structure. The final sample included 140 human trafficking cases.
Text adapted from Farrell, Owens & McDevitt (2013)
Once we obtain G[ν] , given a wind velocity vector u , we can calculate a theoretical R[ν] according to (33). By applying the DFT to R[ν] following (41) in Appendix A, we obtain a theoretical spectrum curve R[κ] . Let Robs[κ] be a spectrum calculated by (41) or (46) from observed data. By comparing R[κ] and Robs[κ] , we can evaluate how close the theoretical spectrum is to the one observed. One of the most popular ways of evaluating the goodness of fit of R[κ] is the least mean squared (LMS) method in which the squared sum of the residue between the two is evaluated.
Another evaluation method is the maximum likelihood (ML), which pursues the maximization of the likelihood as a function R[κ] with respect to Robs[κ] , considering every R[κ] at an integer κ as a scalar random variable that follows a class of gamma distribution, which a sum of squares of Gaussian random variables obey (often mentioned as “χ2 distribution” in a not so rigorous situation). The ML technique is detailed, for example, by Rice . In this article, we employ the ML method to demonstrate our proposed technique.
Text adapted from Nishimura et al. (2020)