diff --git a/text/2_relwork.tex b/text/2_relwork.tex
index 6717c4f..32269e4 100644
--- a/text/2_relwork.tex
+++ b/text/2_relwork.tex
@@ -205,7 +205,7 @@ Quality also depends on the type of platform. \cite{lin2017better} showed that e
 % A comprehensive survey and classification of approaches for community question answering \cite{srba2016comprehensive}, meta study on papers published between 2005 and 2014
 
 
-\subsection{Analysis}
+\section{Analysis}
 
 %general blabla
 % sentiment intensity (Valence based), lexical features
@@ -215,21 +215,27 @@ Quality also depends on the type of platform. \cite{lin2017better} showed that e
 
 % sentiment analyse: es gibt 10-15 methoden, 
 % alle sentiment methoden + vader
-\subsubsection{Sentiment analysis}
+\subsection{Sentiment analysis}
 
 %challenges (vader)
 % - coverage (e.g. of lexical features, important in mircoblog texts)
 % - sentiment intensity (some of the following tools ignore intensity completly (just -1, or 1)
 % - creating a human-validated gold standard lexicon is very time consuming/labor intensive, with sentiment valence scores, feature detection and context awareness, 
 
-%%%%% handcrafted
+
+% polarity-based -> binary
+% valence-base -> continuous
+
+%%%%% handcrafted - TODO order by sofistication, sentwordnet last
 %liwc (Linguistic Inquiry and Word Count) \cite{pennebaker2001linguistic,pennebakerdevelopment}, 2001
-% - acronyms, initialisms, emoticons, or slang, which are known to be important for sentiment analysis of social text (vader)
+% - well verified
+% - ignores acronyms, initialisms, emoticons, or slang, which are known to be important for sentiment analysis of social text (vader)
 % - cannot recognise sentiment intensity (all word have an equal weight) (vader)
 % - ca 4500 words (uptodate?), ca 400 pos words, ca 500 neg words, lexicon proprietary (vader)
+% - TODO list some application examples 
 % ...
 %General Inquirer (GI) \cite{stone1966general} 1966
-% - 11k words, 1900 pos, 2300 neg, all approx
+% - 11k words, 1900 pos, 2300 neg, all approx (vader)
 % - very old (1966), continuously refined, still in use (vader)
 % - misses lexical feature detection (acronyms, ...) and sentiment intensity (vader)
 %Hu-Liu04 \cite{hu2004mining,liu2005opinion}, 2004
@@ -238,18 +244,35 @@ Quality also depends on the type of platform. \cite{lin2017better} showed that e
 % - bootstrapped from wordnet (wellknown english lexical database) (vader)
 %Word-Sense Disambiguation (WSD) \cite{akkaya2009subjectivity}, 2009
 % - TODO
+% - not a sentiment analysis tool per se but can be combined with sentiement analysis tool to distinuish multiple meaning for a word (vader)
+% - a word can have multiple meanings, pos neu neg depending on context (vader)
+% - derive meaning from context -> disambiguation (vader)
+%ANEW (Affective Norms for English Words) \cite{bradley1999affective} 1999
+% - lexicon: 1034 words, ranked by pleasure, arousal, and dominance (vader)
+% - words get value 1-9 (neg-pos, continuous), 5 neutral (TODO maybe list word examples with associated value) (vader)
+% - therefore captures sentiement intensity (vader)
+% - misses lexical features (e.g. acronyms, ...) (vader)
+%SenticNet \cite{cambria2010senticnet} 2010
+% - concept-level opinion and sentiment analysis tool (vader)
+% - sentic mining: combination of AI and Semantic Web (vader)
+% - graphmining and dimensionality reduction (vader)
+% - lexicon: 14250 common-sense concepts, with polarity scores [-1,1] continuous, and many other values (vader)
+% - TODO list some concepts (vader)
 %wordnet \cite{miller1998wordnet} 1998
+% - well-known English lexical database (vader)
+% - group synonyms (synsets) together (vader)
 % - TODO
 %sentiwordnet \cite{baccianella2010sentiwordnet}
-% - TODO
-%ANEW (Affective Norms for English Words) \cite{bradley1999affective}
-% - TODO
-%SenticNet \cite{cambria2010senticnet}
-% - TODO
+% - extension of wordnet (vader)
+% - 147k synset, with 3 values for pos neu neg, sum of synset (TODO pos neu neg?) = 1, range 0-1 continuous (vader)
+% - synset values calc by complex mix of semi supervised algorithms (properagtion methods and classifiers) -> not a gold standard lexicon (vader)
+% - lexicon very noisy, most synset not pos or neg but mix (vader)
+% - misses lexical features (vader)
 
 %%%%% automated (machine learning)
 %often require large training sets, compare to creating a lexicon  (vader)
 %training data must represent as many features as possible, otherwise feature is not learned, often not the case  (vader)
+%training data should be unbiased, or else wrong learning (NOT VADER)
 %very cpu and memory intensive, slow, compare to lexicon-based  (vader)
 %derived features not nachvollziehbar as a human (black-box)  (vader)
 %generaization problem (vader)
@@ -283,7 +306,7 @@ Quality also depends on the type of platform. \cite{lin2017better} showed that e
 
 % its 
 % ursprüngliches paper ITS, wie hat man das früher (davor) gemacht
-\subsubsection{Trend analysis}
+\subsection{Trend analysis}