In our previous works [1][2][3][4][5][6][7] we used both symmetrical (no diquark) core quark distribution and asymmetrical (diquark-quark) core quark distribution along with pseudoscalar mesons and vector mesons to calculate polarized [1,5,6] and unpolarized [2][3][4]7] nucleon structure functions. For the asymmetrical distribution we considered the superposition of spin-0, isospin-0 and spin-0, isospin-1 along with spin-0, isospin-0 and spin-1 isospin-1 states. It turned out that for unpolarized structure functions only the quark-diquark model was able to reproduce experimental results reasonably well [2][3][4]7]. Also, one needs the contribution of vector mesons to reproduce the observed Gottfried sum-rule violation [13]. However, for the polarized case the results were rather mixed. For some case the symmetrical model was more agreeable with observation [1,5,6]. Our objective here is to have a core quark distribution that can reproduce observation equally well, whether it is for the polarized structure function or the unpolarized structure function. To achieve this goal we have added the other two possible states to the diquark state. Namely, spin-1, isospin-0 and spin-1, isospin-1 and the contribution of both pseudoscalar mesons and vector mesons. The calculation is performed in the light-cone frame. The dressed nucleon is assumed to be a superposition of the bare nucleon plus virtual light-cone Fock states of baryon-meson pairs. For bare nucleon we consider both the case of diquark-quark model which is now the superposition of all four diquark states and the case which there is no quark clustering inside the nucleon. The initial distributions are evolved using DGLAP equations. The final results are compared with experimental results and other theoretical predictions.