Minor presentation fixes

appendices/a_captions.md

@@ -62,7 +62,7 @@ source.
 ![Fig. 4d reproduced from the paper, a scammer profile image][fig4d]
 
 1. A man sitting in front of a laptop computer (`p = 13.1e-03`)
-2. A man sitting at a table with a laptop (`p = 3.6e-03)
+2. A man sitting at a table with a laptop (`p = 3.6e-03`)
 3. A man sitting at a table with a laptop computer (`p = 2e-03`)
 
 These descriptions belong to an image from the _scammer_ category. The image 

appendices/b_topics.md

@@ -3,10 +3,10 @@
 [fig7]: figures/descr_topics.png
 
 As mentioned in Section 2.3 of the paper, we used the dictionary terms that are
-mapped to categories from the [LIWC 2015 dictionary](https://
-www.liwc.net/dictionaries/) to analyse the variety of topics that are present in
-both scammer and real profile descriptions. The results of the analysis are
-shown in Fig. 7 below.
+mapped to categories from the [LIWC 2015
+dictionary](https://www.liwc.net/dictionaries/) to analyse the variety of topics
+that are present in both scammer and real profile descriptions. The results of
+the analysis are shown in Fig. 7 below.
 
 ![Fig. 7: topic variety in user profile descriptions based on the LIWC 2015
 dictionary][fig7]

appendices/d_elements.md

@@ -1,21 +1,21 @@
 # Appendix D: Comparison of Dating Site Elements
 
 
-The [datingnmore.com] site used as the source of data for our experiments is
-comparatively small, and has a niche appeal due to its intensive moderation by
-experts in identifying online dating fraud.  It is therefore worthwhile
-considering its comparability to other dating sites, as a first step to
-understand the generalisability of our results.
+The [datingnmore.com](https://datingnmore.com) site used as the source of data
+for our experiments is comparatively small, and has a niche appeal due to its
+intensive moderation by experts in identifying online dating fraud.  It is
+therefore worthwhile considering its comparability to other dating sites, as a
+first step to understand the generalisability of our results.
 
-Table VII compares the features from the [datingnmore.com] profiles which were
-used in our classifier with the profile elements available on five
-market-leading dating sites.  Coverage is good. All three of the ensemble
-components would be able to operate across these sites. The image and
-description profile elements are always supported, and at least some demographic
-information is always available. The dating site with the fewest profile
-elements in common with our features is [tinder.com], which has a distinctive
-locality-based use case which may hinder the form of international online dating
-fraud our system aims to detect.
+Table VII compares the features from the
+[datingnmore.com](https://datingnmore.com) profiles which were used in our
+classifier with the profile elements available on five market-leading dating
+sites.  Coverage is good. All three of the ensemble components would be able to
+operate across these sites. The image and description profile elements are
+always supported, and at least some demographic information is always available.
+The dating site with the fewest profile elements in common with our features is
+[tinder.com], which has a distinctive locality-based use case which may hinder
+the form of international online dating fraud our system aims to detect.
 
 ### Table VII: Comparison of the profile elements used in our classification experiments with availability of these elements on popular dating sites.  ( *✔* :present; *%*: requires inference)
 
@@ -27,3 +27,9 @@ fraud our system aims to detect.
 |[pof.com] 	  | *✔* | *✔*	 | *✔*	 | 	*✔*|*✔*   | *✔*	     |*✔*    | *✔*    |
 |[eharmony.com]   | *✔* | *✔*	 | *✔*	 | 	*✔*|*✔*   | *✔*	     |*✔*    | *✔*    |
 |[tinder.com] 	  | *✔* | *✔*	 | 	 | 	   |      | *%*	     |*✔*    | *✔*    |
+
+[match.com]: https://match.com 
+[okcupid.com]: https://okcupid.com 
+[pof.com]: https://pof.com 
+[eharmony.com]: https://eharmony.com 
+[tinder.com]: https://tinder.com