Ignoring 2008 for the moment, UCR
Expanded Homicide Data Table 9 says for 2007 there were
14,831 total murder victims
of which
1,924 were killed by a stranger and
6,848 were killed by someone, relationship unknown.
I don't think 50% by strangers can be supported.
OTOH, ordinarily family and acquaintances are initially 'good suspects' and often are investigated; if the investigation is thorough, and those initial suspects are not confirmed, 'stranger' is what's left, mixed with 'we are sure X did it, but cannot prove it', in unknown proportion.
As to the distinction between 'homicide' and 'murder', it's certainly significant in a legal sense, but not particularly important in a statistical sense.
FBI says
Law enforcement reported 645 justifiable homicides in 2007. Of those, law enforcement officers justifiably killed 391 individuals, and private citizens justifiably killed 254 individuals.
645 is just over 4% of the murders+justifiable homicides.
Similarly,
Law enforcement reported 617 justifiable homicides in 2006. Of those, law enforcement officers justifiably killed 376 individuals, and private citizens justifiably killed 241 individuals.
17,034 murders reported, so 617 was about 3.5% of total homicides.
Odds of being murdered, of course are strongly related to race, gender and age. In 2005, black males 20-24 years old were victims of homicide at 111.55 / 100,000, 25-29 years - 100.55 / 100,000.
Frankly, that's terrifying.
ETA -- and 2005 was a
good year - for the same two groups, the rates per 100,000 were 1989: 134 and 131; 1990: 141 and 136; 1991: 154 and 133; 1992: 147 and 122; 1993: 163 and 124.