include mroe words in the table

analyses/category_prediction.py

@@ -172,7 +172,7 @@ if __name__ == '__main__':
 
     score_vs_vocab_size = []
 
-    for vocab in np.arange(125, 1000+125, 125):
+    for vocab in [50] + list(np.arange(125, 1000+125, 125)):
         score = 0
         for i in range(3):
             dummies[i] = DummyClassifier(strategy="most_frequent")
@@ -316,8 +316,8 @@ if __name__ == '__main__':
 
     cat = "th"
 
-    top = results[results["Phenomenology-HEP"]>0].sort_values(f"ph_minus_{cat}", ascending=False).head(30).index.values
-    bottom = results[results[cats[cat]]>0].sort_values(f"ph_minus_{cat}", ascending=True).head(30).index.values
+    top = results[results["Phenomenology-HEP"]>0].sort_values(f"ph_minus_{cat}", ascending=False).head(40).index.values
+    bottom = results[results[cats[cat]]>0].sort_values(f"ph_minus_{cat}", ascending=True).head(40).index.values
         
     table.append({
         'Reference category': cats_friendly[cat],

tables/specific_vocabulary.tex

@@ -2,12 +2,12 @@
 \caption{Vocabulary specific to each category. The left column lists expressions that discriminate experiment and theory from phenomenology. The right column lists expressions that are the most specific to phenomenology and foreigh to experiment and theory.}
 \label{table:specific_pheno_vocabulary}\\
 \toprule
-{} &                                                                                                                                                                                                                                                                                                                                                                                                                 Vocabulary distinct from phenomenology &                                                                                                                                                                                                                                                                                                                                                                                                                                                Vocabulary specific to phenomenology \\
+{} &                                                                                                                                                                                                                                                                                                                                                                                                                                                  Vocabulary specific to phenomenology &                                                                                                                                                                                                                                                                                                                                                                                                                    Vocabulary specific to theory \\
 \midrule
 \endfirsthead
 \caption[]{Vocabulary specific to each category. The left column lists expressions that discriminate experiment and theory from phenomenology. The right column lists expressions that are the most specific to phenomenology and foreigh to experiment and theory.} \\
 \toprule
-{} &                                                                                                                                                                                                                                                                                                                                                                                                                 Vocabulary distinct from phenomenology &                                                                                                                                                                                                                                                                                                                                                                                                                                                Vocabulary specific to phenomenology \\
+{} &                                                                                                                                                                                                                                                                                                                                                                                                                                                  Vocabulary specific to phenomenology &                                                                                                                                                                                                                                                                                                                                                                                                                    Vocabulary specific to theory \\
 \midrule
 \endhead
 \midrule
@@ -17,6 +17,6 @@
 
 \bottomrule
 \endlastfoot
-\textbf{Experiment} &                                                              detector, sample, measurement, search, confidence, experiment, upper limit, target, beam, cm, result, evidence, luminosity, proton-proton collision, fraction, gev, tev, event, first time, resolution, limit, expectation, top quark, branching, data, recent result, muon, -1, construction, yield, mev, status, system, technique, world, study, product, final state, range, electron &                          qcd, experimental data, quark, heavy ion collision, mssm, dark matter, color, high energy, parton, constraint, factorization, lhc, gluon, contribution, possibility, process, chiral symmetry, plasma, calculation, electroweak, neutrino, effect, account, evolution, phenomenology, implication, new physic, mixing, scenario, higgs, unification, framework, model, mass, violation, estimate, medium, direct detection, extra dimension, approximation \\
-\textbf{Theory    } &  algebra, spacetime, manifold, theory, gravity, branes, action, partition, chern-simons, string, geometry, supergravity, ad, deformation, space-time, horizon, quantum, central charge, duality, entropy, conformal field, coordinate, modulus space, construction, yang-mills, correspondence, surface, space, transformation, dimension, sitter, formulation, conformal field theory, conjecture, hole, solution, black hole, field, massless, class &  quark, qcd, neutrino, lhc, extra dimension, parton, dark matter, electroweak, phenomenology, experimental data, mssm, high energy, collider, nucleon, sensitivity, effective theory, hadron, standard model, gluon, color, chiral perturbation, new physic, chiral perturbation theory, neutron, contribution, neutrino mass, process, pion, effective field, effective field theory, sm, channel, flavor, decay, distribution, signal, exchange, next-to-leading order, cp, meson \\
+\textbf{Experiment} &  quark, experimental data, qcd, mssm, dark matter, lhc, color, high energy, factorization, heavy ion collision, new physic, effect, constraint, evolution, gluon, potential, account, mixing, numerical result, generator, parton, implication, contribution, inflation, higgs, next-to-leading order, impact, estimate, neutrino, calculation, unification, approximation, neutrino mass, scenario, process, bound, chiral perturbation theory, direct detection, possibility, model &                                                          detector, sample, measurement, search, experiment, upper limit, confidence, atlas, proton-proton collision, target, event, gev, beam, evidence, result, fraction, luminosity, branching, -1, recent result, resolution, first time, tev, top quark, technique, limit, article, cm, expectation, study, mev, yield, muon, ii, reaction, final state, invariant mass, world, data, system \\
+\textbf{Theory    } &                 quark, lhc, neutrino, qcd, dark matter, standard model, extra dimension, experimental data, mssm, gluon, mixing, next-to-leading order, effective theory, nucleon, collider, high energy, chiral perturbation theory, chiral perturbation, electroweak, process, impact, hadron, neutrino mass, distribution, color, parton, moment, phenomenology, baryon, factorization, meson, asymmetry, decay, contribution, sm, production, new physic, cp, exchange, violation &  algebra, manifold, geometry, spacetime, chern-simons, ad, theory, gravity, action, space-time, quantum, yang-mills, supergravity, duality, string, partition, correspondence, deformation, sitter, branes, modulus space, surface, construction, central charge, space, cosmological constant, horizon, dimension, conformal field, field, entropy, coordinate, path, hole, brane, black hole, solution, conformal field theory, boundary, type \\
 \end{longtable}

tables/specific_vocabulary_th_ph.tex

@@ -2,12 +2,12 @@
 \caption{Vocabulary specific to theory (left column) versus phenomenology (right column). }
 \label{table:specific_pheno_vocabulary_th_ph}\\
 \toprule
-                                                                                                                                                                                                                                                                                                  Vocabulary specific to theory &                                                                                                                                                                                                                                                                                                                                                 Vocabulary specific to phenomenology \\
+                                                                                                                                                                                                                                                                                                                                                                                                                                 Vocabulary specific to phenomenology &                                                                                                                                                                                                                                                                                                                                                                                                                   Vocabulary specific to theory \\
 \midrule
 \endfirsthead
 \caption[]{Vocabulary specific to theory (left column) versus phenomenology (right column). } \\
 \toprule
-                                                                                                                                                                                                                                                                                                  Vocabulary specific to theory &                                                                                                                                                                                                                                                                                                                                                 Vocabulary specific to phenomenology \\
+                                                                                                                                                                                                                                                                                                                                                                                                                                 Vocabulary specific to phenomenology &                                                                                                                                                                                                                                                                                                                                                                                                                   Vocabulary specific to theory \\
 \midrule
 \endhead
 \midrule
@@ -17,5 +17,5 @@
 
 \bottomrule
 \endlastfoot
-algebra, spacetime, manifold, theory, gravity, branes, action, partition, chern-simons, string, geometry, supergravity, ad, deformation, space-time, horizon, quantum, central charge, duality, entropy, conformal field, coordinate, modulus space, construction, yang-mills, correspondence, surface, space, transformation, dimension & quark, qcd, neutrino, lhc, extra dimension, parton, dark matter, electroweak, phenomenology, experimental data, mssm, high energy, collider, nucleon, sensitivity, effective theory, hadron, standard model, gluon, color, chiral perturbation, new physic, chiral perturbation theory, neutron, contribution, neutrino mass, process, pion, effective field, effective field theory \\
+quark, lhc, neutrino, qcd, dark matter, standard model, extra dimension, experimental data, mssm, gluon, mixing, next-to-leading order, effective theory, nucleon, collider, high energy, chiral perturbation theory, chiral perturbation, electroweak, process, impact, hadron, neutrino mass, distribution, color, parton, moment, phenomenology, baryon, factorization, meson, asymmetry, decay, contribution, sm, production, new physic, cp, exchange, violation & algebra, manifold, geometry, spacetime, chern-simons, ad, theory, gravity, action, space-time, quantum, yang-mills, supergravity, duality, string, partition, correspondence, deformation, sitter, branes, modulus space, surface, construction, central charge, space, cosmological constant, horizon, dimension, conformal field, field, entropy, coordinate, path, hole, brane, black hole, solution, conformal field theory, boundary, type \\
 \end{longtable}